This overview of 15 apps helps compare transcription of recordings, meeting notes and offline tools. We do not name an accuracy winner: without a shared recording benchmark, that would not be fair. The choice depends on conversation language, hardware, privacy, limits and cost.
The author of this comparison is the creator of Vocatim, one of the products scored. Scores are editorial, not user reviews. Documentation checked: October 10, 2026; Vocatim: version 1.7.0. Prices refer to the named offers, not identical feature packages.

Vocatim
4.4 / 5 · 12/12 categories
App website · App Store
Baseline average, without priorities, using a scenario exchange rate of 1.10 USD/EUR. Average range: 4.3–4.6. Interactive values may change when weights are selected.
- English transcription
- Yes · Whisper; depends on the selected engine / mode · Source
- English interface
- Yes
- Who it suits
- Private conversations and notes on iPhone and Mac.
- Main limitation
- No Android or Windows app. Translation requires OS 26.4+; performance depends on the device.
- Availability
- Also iPad. Mac with Apple Silicon. One purchase for Apple devices.
- Transcription languages
- 100 · Whisper large-v3 through Auto. 18 interface languages. Whisper recognizes 100 languages in Auto mode; 18 are also available for manual selection. Live preview, translation and summaries have separate language lists.
- Interface
- 18 localizations · Vocatim 1.7.0 · iPhone / Mac. 18 UI_LANGUAGES entries in the code for release 1.7.0, commit efc6e21b. These are menu languages, not a limit on speech recognition. We did not assess the quality of each translation.
- Features
- Local translation after downloading languages; iOS/iPadOS/macOS 26.4+.
- Live
- Draft transcription, notes that develop during recording, and translated captions. Captions have a delay; this is not simultaneous interpreting. Live features support fewer languages than file transcription.
- Cloud models
- No external STT in the described 1.7 scope. No external LLM in the described 1.7 scope.
- Recordings and text
- Recordings: On the device; optionally in your own iCloud or on a paired Mac. Transcripts: On the device; optional transfers through your own iCloud.
- Processing and training
- On iPhone / Mac; optionally on your own paired Mac, not on Vocatim servers. Local Qwen3.5 and Apple translation. Your own iCloud handles transfer / sync, not AI inference. The developer does not receive recording content
- Reference plan
- Full version · one-time purchase: $24.99 in the first year; $0.00 in the following year at the same prices. US price. Germany: EUR 29.99. Local prices may vary.
- Average annual cost over 2 years
- $12.50
- Features for the price
- Vocatim 1.7.0, full Apple license. Translation requires a compatible operating system and downloaded languages.
- Free option
- Ongoing preview · first 5 min of each recording. Local transcription of the first 5 minutes of any recording. No single exhaustible trial allowance. The portion of the transcript already available can be copied and exported. The preview covers only the first 5 min of the recording, not the full 30 min. Export does not unlock the omitted portion or establish free access to all AI features. Version 1.7.0 scope. Segment copying and TXT/SRT/VTT/DOCX export confirmed by reviewing this version of the code: no purchase gate in the menus or handlers. This is code verification, not an on-device purchase test.
- Meetings
- Microphone; on Mac also system audio + microphone. No bot.
- Files and export
- Audio import; TXT, DOCX, SRT, VTT plus Markdown and JSON in 1.7. Export the original, translation, or both. Automation through Apple Shortcuts / App Intents.
- Teamwork
- A personal tool. Syncing your own devices is not a shared team workspace or SSO administration. Shortcuts enable custom workflows but do not replace native CRM integrations.
- Mobile models
- Whisper base / large-v3 / large-v3-turbo. Parakeet v3 and Nemotron are selected according to language, rather than being five equivalent choices. Separate speaker and embedding models.
- Automation
- Apple Shortcuts / App Intents: file import for transcription, latest transcript and summary.
Category scores for Vocatim| Category | Score |
|---|
| Availability | 3 / 5 |
|---|
| Internationalization | 4.6 / 5 |
|---|
| AI models | 5 / 5 |
|---|
| Functionality | 5 / 5 |
|---|
| Live features | 5 / 5 |
|---|
| Privacy / security | 5 / 5 |
|---|
| Cost | 4 / 5 |
|---|
| Files and export | 5 / 5 |
|---|
| Meetings and collaboration | 3–4 / 5 |
|---|
| Agents and automation | 3–5 / 5 |
|---|
| Free mode | 4 / 5 |
|---|
| Features for the price | 4.6 / 5 |
|---|
Manufacturer sources

Granola
3.6 / 5 · 12/12 categories
App website · App Store
Baseline average, without priorities, using a scenario exchange rate of 1.10 USD/EUR. Average range: 3.4–3.9. Interactive values may change when weights are selected.
- English transcription
- Yes · Source
- Who it suits
- AI-enhanced notes, without a bot joining the conversation.
- Main limitation
- Cloud processing. No audio archive for later playback.
- Availability
- Mac, Windows, iPhone, Android; also Apple Watch. Features vary between clients.
- Transcription languages
- 31 desktop / 17 mobile. Current help distinguishes 31 languages on Mac/Windows from 17 on iOS/Android. Polish is on the mobile list but not the desktop list. We count the broader single client, not the union of both lists. Multilingual transcription on desktop; mobile selects one language.
- Interface
- 1 localizations · Desktop and mobile. Official help confirms that menus, buttons and settings are English-only.
- Features
- Speaker tags in Zoom / Meet / Teams on desktop and speaker separation on mobile. Quote translation through Chat, not full transcript localization. Persistent voice recognition remains unconfirmed.
- Live
- Desktop transcribes during the meeting. Chat can summarize the conversation so far and decisions on request, so live summarization is partial. These are not automatically refreshed final notes. Live translation unconfirmed; mobile may process after recording.
- Cloud models
- 2 disclosed providers: Deepgram and AssemblyAI. Model counts and versions undisclosed. Business / Enterprise: choice of standard and reasoning models; Basic: Auto. The help center mentions Fable 5, but not the full catalog. Fable 5 has 30-day provider retention, not ZDR.
- Recordings and text
- Recordings: No permanent archive; mobile temporarily buffers audio before transcription.. Transcripts: Cloud, AWS infrastructure in the US.
- Processing and training
- Cloud STT: Deepgram / AssemblyAI. Desktop during the meeting; mobile from a temporary audio buffer. Cloud: OpenAI / Anthropic models used for notes and AI. In-house training on data described as anonymous; opt-out available
- Reference plan
- Business · 12 × USD 14: $168.00 in the first year; $168.00 in the following year at the same prices. The cost of 12 monthly payments, not an annual billing discount.
- Average annual cost over 2 years
- $168.00
- Features for the price
- Business. Translation through chat; on-demand summaries during meetings. The full feature set differs between desktop and mobile.
- Free option
- Basic · ongoing free plan. Transcriptions with no daily limit, AI notes and Chat (Auto), folders, templates, Slack. Text copying and CSV export of your own notes. In-app history: 30 days. CSV also includes older notes with summaries; link valid for 24 h, one export request per day. Maximum Basic transcription length unconfirmed. CSV available in Basic by default; integrations beyond Slack and model selection require a paid plan. Older notes are stored, not deleted after 30 days.
- Meetings
- Bot-free device audio capture; no audio archive.
- Files and export
- No import of previously recorded audio files. Copy transcripts and export your own notes with summaries as CSV; 1 export per 24 h. Zapier on Business / Enterprise.
- Teamwork
- Shared spaces, integrations and administration; some controls require Enterprise.
- Mobile models
- Remote and local models are treated equally. Model selectors and specific versions depend on the client and plan.
- Automation
- Zapier; integrations include Notion, Slack and CRM.
Category scores for Granola| Category | Score |
|---|
| Availability | 5 / 5 |
|---|
| Internationalization | 2.2 / 5 |
|---|
| AI models | 3.5–5 / 5 |
|---|
| Functionality | 4–5 / 5 |
|---|
| Live features | 3–4 / 5 |
|---|
| Privacy / security | 1 / 5 |
|---|
| Cost | 2 / 5 |
|---|
| Files and export | 4 / 5 |
|---|
| Meetings and collaboration | 5 / 5 |
|---|
| Agents and automation | 5 / 5 |
|---|
| Free mode | 4–5 / 5 |
|---|
| Features for the price | 2–2.3 / 5 |
|---|
Manufacturer sources

Otter
3.9 / 5 · 12/12 categories
App website
Baseline average, without priorities, using a scenario exchange rate of 1.10 USD/EUR. Average range: 3.7–4.1. Interactive values may change when weights are selected.
- English transcription
- Yes · Source
- Who it suits
- Team meetings, shared notes and AI Chat.
- Main limitation
- Polish is not on the official list. Pro has a separate imported-file limit.
- Availability
- Mobile apps, desktop and browser; features depend on the client.
- Transcription languages
- 6. 6 transcription languages: EN, ES, FR, DE, JA, ZH. The interface remains in English. Translation through AI Chat does not expand the STT list.
- Interface
- 1 localizations · Web and mobile. Otter help clearly distinguishes its 6 transcription languages from menus and settings, which remain in English.
- Features
- Otter learns labeled voices across conversations; Enterprise can disable this. Translation through Otter Chat is officially documented: partial credit for a response to a prompt, not a dedicated bilingual export or a guarantee of live translation. The STT language count remains unchanged.
- Live
- Transcription and summaries during the conversation. The full scope of live translation has not been confirmed.
- Cloud models
- Otter’s proprietary STT technology; model counts and versions undisclosed. In-house models and several external providers. The FAQ does not name versions.
- Recordings and text
- Recordings: Cloud, AWS US. Transcripts: Cloud, AWS US.
- Processing and training
- Otter’s own cloud speech recognition. AI Chat: in-house LLMs and external providers. Otter describes training on content with identifiers removed
- Reference plan
- Pro · annual billing: $99.96 in the first year; $99.96 in the following year at the same prices. About USD 100: 12 × the rounded USD 8.33. The final amount may differ slightly.
- Average annual cost over 2 years
- $99.96
- Features for the price
- Pro. Translation through AI Chat. Query and import limits still apply; the Enterprise API is not included.
- Free option
- Basic · 300 min/month. Live transcription, speaker recognition, playback and AI chat; TXT text export. 30 min per conversation; 3 file imports for the lifetime of the account, not per month. History of the 25 most recent conversations. The monthly allowance applies to recording; the import limit is separate. Additional export formats and features require payment.
- Meetings
- Meeting assistant and recording; audio archive for playback.
- Files and export
- Basic: TXT and 3 imports for the lifetime of the account. Paid plans: DOCX, PDF and SRT; bulk export from Business. Pro: 10 imports/month. Automatic API import requires Enterprise, not the reference Pro plan.
- Teamwork
- Shared notes and integrations. CRM, advanced roles and SSO depend on a higher plan.
- Mobile models
- Remote and local models are treated equally. Model selectors and specific versions depend on the client and plan.
- Automation
- Zapier, Salesforce and HubSpot; scope depends on the plan.
Category scores for Otter| Category | Score |
|---|
| Availability | 5 / 5 |
|---|
| Internationalization | 1 / 5 |
|---|
| AI models | 3–5 / 5 |
|---|
| Functionality | 5 / 5 |
|---|
| Live features | 4–5 / 5 |
|---|
| Privacy / security | 1 / 5 |
|---|
| Cost | 3 / 5 |
|---|
| Files and export | 5 / 5 |
|---|
| Meetings and collaboration | 5 / 5 |
|---|
| Agents and automation | 5 / 5 |
|---|
| Free mode | 5 / 5 |
|---|
| Features for the price | 2.7–2.9 / 5 |
|---|
Manufacturer sources

MacWhisper
4.0 / 5 · 12/12 categories
App website · App Store
Baseline average, without priorities, using a scenario exchange rate of 1.10 USD/EUR. Average range: 3.7–4.4. Interactive values may change when weights are selected.
- English transcription
- Yes · Whisper; depends on the selected engine / mode · Source
- Who it suits
- Files, automation and a wide choice of models on Mac.
- Main limitation
- Separate iOS offering. Cloud APIs and Assistant may cost extra.
- Availability
- iPhone/iPad app named Whisper Transcription. A MacWhisper Pro license is a different purchase from the App Store offering.
- Transcription languages
- 100+. The count applies to the Whisper offering. A specific engine, such as Apple Speech or Parakeet, has a different language list.
- Interface
- 4 localizations · Whisper Transcription · App Store, iPhone / Mac. Localization list declared by the vendor in the App Store, checked on October 10, 2026. This is not an audit of translation completeness. The count applies to the App Store version. The complete list for MacWhisper downloaded directly from the vendor is unconfirmed; the price in this comparison is for that direct license.
- Features
- AI can run locally with additional setup or through APIs. DeepL translation sends text to DeepL; not all features are local.
- Live
- Live dictation confirmed. We do not equate it with continuous transcription of a meeting with several participants and a live summary.
- Cloud models
- The official desktop instructions list OpenAI, Groq, ElevenLabs, Deepgram and a custom OpenAI-compatible endpoint. These are four named providers, not four models; exact versions depend on the service. iOS has an optional Assistant. Choices include OpenAI, Claude, Gemini, Grok, DeepSeek, OpenRouter and Groq; bring your own API keys. Model versions / counts vary. Assistant is a separate subscription for the App Store version.
- Recordings and text
- Recordings: Local with a local model; optional cloud provider. Transcripts: Local; sent to external AI or DeepL when those features are selected.
- Processing and training
- Local STT or selected external APIs, depending on the engine. Locally through Ollama / LM Studio or external AI; DeepL processes submitted text. Depends on the selected provider; local mode does not send content
- Reference plan
- Pro for Mac · one-time purchase: €64.00 in the first year; €0.00 in the following year at the same prices. Direct Mac license. Prices are compared in USD only using the disclosed, editable scenario exchange rate.
- Average annual cost over 2 years
- $35.20
- Features for the price
- Pro on Mac, excluding Assistant and paid APIs. Summaries/chat with a free local LLM require setup. Whisper translates into English; other workflows may cost extra.
- Free option
- MacWhisper Free · ongoing plan on Mac. Microphone recording and file transcription; TXT, SRT and VTT. Pricing distinguishes Free from Pro. TXT/SRT/VTT/CSV through the CLI do not require Pro; JSON/Markdown/HTML/Avid require Pro. Recording length and a recurring Free transcription allowance on Mac remain unconfirmed; the phrase Free forever is not taken to mean unlimited use. Free scoring applies to the direct Mac version. iOS offers free local models and an optional Assistant, but these have separate terms, which are not applied to Mac.
- Meetings
- Mac: device audio and microphone recording; meeting detection in beta.
- Files and export
- Batches, watched folders, captions, documents, CLI and integrations. A much broader toolkit than a simple voice recorder.
- Teamwork
- MDM, volume licenses and integrations; these are not the same as a shared cloud editor with roles.
- Mobile models
- Whisper Transcription on iOS lets you choose local STT or the cloud Assistant. The transcription method/engine choice is confirmed, but not the full model count or an LLM selector. Local mode requirements depend on the device.
- Automation
- Scripts, pipelines and watched folders.
Category scores for MacWhisper| Category | Score |
|---|
| Availability | 3 / 5 |
|---|
| Internationalization | 3.8 / 5 |
|---|
| AI models | 4.5–5 / 5 |
|---|
| Functionality | 4–5 / 5 |
|---|
| Live features | 2–4 / 5 |
|---|
| Privacy / security | 4 / 5 |
|---|
| Cost | 4 / 5 |
|---|
| Files and export | 5 / 5 |
|---|
| Meetings and collaboration | 4–5 / 5 |
|---|
| Agents and automation | 4–5 / 5 |
|---|
| Free mode | 3–5 / 5 |
|---|
| Features for the price | 2.8–3.7 / 5 |
|---|
Manufacturer sources

Aiko
2.9 / 5 · 12/12 categories
App website · App Store
Baseline average, without priorities, using a scenario exchange rate of 1.10 USD/EUR. Average range: 2.7–2.9. Interactive values may change when weights are selected.
- English transcription
- Yes · Source
- Who it suits
- Simple local audio-to-text conversion in many languages.
- Main limitation
- No diarization, text editing or AI notebook features.
- Availability
- iPhone, iPad, Mac, visionOS. The current iPhone version requires iOS 26.
- Transcription languages
- 100. Multilingual transcription does not mean a multilingual interface. Whisper translation is into English only.
- Interface
- 1 localizations · Aiko · iPhone / Mac. Localization list declared by the vendor in the App Store, checked on October 10, 2026. This is not an audit of translation completeness.
- Features
- Translation into English, not any language pair. The FAQ confirms no speaker separation or editor.
- Live
- Import and recording for later transcription. The developer explicitly rules out live transcription.
- Cloud models
- No cloud STT in the described app. No built-in LLM; a custom shortcut to external AI is a separate data flow.
- Recordings and text
- Recordings: Locally on the device. Transcripts: Locally on the device.
- Processing and training
- Local Whisper on the device. Local STT output; no in-house LLM for notes. The developer states that no app data is collected
- Reference plan
- One-time purchase · Apple: $24.00 in the first year; $0.00 in the following year at the same prices. The current offering is not free. We do not compare prices of historical versions.
- Average annual cost over 2 years
- $12.00
- Features for the price
- Apple app purchase. Translation into English; no AI notebook or live mode.
- Free option
- 14-day trial · TestFlight. The full app with no feature restrictions during the trial. Local transcription of recordings. After 14 days, the current version requires a purchase. This is not an ongoing free plan. Older releases for older macOS versions are free, but they are not the current version compared here.
- Meetings
- Microphone recording and importing existing audio.
- Files and export
- File import and text and SRT export. No built-in editor; make corrections in another app. Official shortcuts also support batches.
- Teamwork
- Personal app with export, not a shared meeting workspace.
- Mobile models
- Automatic variant selection is not a full catalog of models for manual selection.
- Automation
- Official Apple Shortcuts: transcription, batches and forwarding results.
Category scores for Aiko| Category | Score |
|---|
| Availability | 3 / 5 |
|---|
| Internationalization | 3.4 / 5 |
|---|
| AI models | 2 / 5 |
|---|
| Functionality | 2 / 5 |
|---|
| Live features | 1 / 5 |
|---|
| Privacy / security | 5 / 5 |
|---|
| Cost | 4 / 5 |
|---|
| Files and export | 5 / 5 |
|---|
| Meetings and collaboration | 2 / 5 |
|---|
| Agents and automation | 3–5 / 5 |
|---|
| Free mode | 1 / 5 |
|---|
| Features for the price | 1.9 / 5 |
|---|
Manufacturer sources

Fireflies
4.0 / 5 · 12/12 categories
App website · App Store
Baseline average, without priorities, using a scenario exchange rate of 1.10 USD/EUR. Average range: 3.8–4.2. Interactive values may change when weights are selected.
- English transcription
- Yes · Source
- Who it suits
- Collaboration, automated notes and integrations with team tools.
- Main limitation
- Storage allowance and imports have separate limits.
- Availability
- Desktop on Mac and Windows; iOS and Android apps and a Chrome extension.
- Transcription languages
- 100+. Claimed transcription coverage. Multilingual conversation mode requires a suitable plan.
- Interface
- 1 localizations · iOS · Fireflies 0.5.55. Localization list declared by the vendor in the App Store, checked on October 10, 2026. This is not an audit of translation completeness. The complete web / desktop interface list is unconfirmed; 100+ refers to transcription.
- Features
- Speaker separation, summaries and AskFred. Some AI features use credits.
- Live
- The desktop description covers transcription, evolving notes and questions during the meeting. Live translation scope not established.
- Cloud models
- STT model and engine count not established. We do not equate AskFred with speech recognition. AskFred: 5 names in help documentation: Claude Sonnet 5, Opus 5, GPT-5.6 Luna, GPT-5.6 Sol, Gemini 3.5 Flash. Auto is not a sixth model.
- Recordings and text
- Recordings: Cloud; configuration-dependent region needs confirmation. Transcripts: Cloud; retention according to account and plan.
- Processing and training
- Fireflies cloud and STT subprocessors. Cloud AI, including models selected in AskFred. Claimed prohibition on training on meeting content
- Reference plan
- Pro · annual billing: $120.00 in the first year; $120.00 in the following year at the same prices. Inconsistent documentation on charges for large imports; we do not give a definitive price for overages.
- Average annual cost over 2 years
- $120.00
- Features for the price
- Pro. AskFred uses credits; extra paid credits and Enterprise features are excluded.
- Free option
- Free · unlimited transcription with Auto-join enabled. Auto-join: unlimited transcripts for eligible meetings; 400 min of storage per seat, 20 AI credits monthly and collaboration. Available transcript text can also be read through the GraphQL API on Free. Imports up to 150 min consume a transcription credit; initial credits depend on signup method. Downloading a transcript using the in-app button requires a paid plan. Free text export is confirmed through the API, not through that button. The length point confirms import of a 150-minute file, not unlimited imports. Ongoing use without an allowance applies to Auto-join. These conditions are not interchangeable. The export point covers technical access through the API: an account key, meeting permissions and rate limits. It does not mean free audio downloads.
- Meetings
- Meeting bot or bot-free desktop audio capture.
- Files and export
- Audio/video import, export and integrations. File and source limits are independent of storage.
- Teamwork
- Sharing, integrations and organization management; enterprise features are not included in the Pro price.
- Mobile models
- Remote and local models are treated equally. Model selectors and specific versions depend on the client and plan.
- Automation
- Integrations and automated post-meeting workflows.
Category scores for Fireflies| Category | Score |
|---|
| Availability | 5 / 5 |
|---|
| Internationalization | 3.4 / 5 |
|---|
| AI models | 3.5–5 / 5 |
|---|
| Functionality | 4–5 / 5 |
|---|
| Live features | 4–5 / 5 |
|---|
| Privacy / security | 2 / 5 |
|---|
| Cost | 2 / 5 |
|---|
| Files and export | 5 / 5 |
|---|
| Meetings and collaboration | 5 / 5 |
|---|
| Agents and automation | 5 / 5 |
|---|
| Free mode | 5 / 5 |
|---|
| Features for the price | 2.3–2.8 / 5 |
|---|
Manufacturer sources

Fathom
3.4 / 5 · 12/12 categories
App website · App Store
Baseline average, without priorities, using a scenario exchange rate of 1.10 USD/EUR. Average range: 3.2–3.7. Interactive values may change when weights are selected.
- English transcription
- Yes · Source
- Who it suits
- A free start with meeting recordings and transcripts.
- Main limitation
- Does not replace an archive file importer; the iOS app requires internet access.
- Availability
- iOS and desktop. Bot-free mode on Mac is labeled beta; we do not assume the same scope on Windows.
- Transcription languages
- 38 claimed. Help documentation claims 38; the list and scope of summary translation are separate. Not all languages have identical features.
- Interface
- 1 localizations · iOS · Fathom Video Inc.. Localization list declared by the vendor in the App Store, checked on October 10, 2026. This is not an audit of translation completeness. This is not a similarly named app from a different vendor. The desktop / web UI language list is unconfirmed.
- Features
- A voice profile for the iOS user does not recognize every speaker. Translated summaries are not a full transcript translation.
- Live
- New bot-free desktop mode: summary updated during the conversation; the help center specifies eligible English-language meetings on Mac. The older bot mode generates results after the meeting. The article does not establish the availability of a full live transcript or live translation.
- Cloud models
- Exact model and model count undisclosed in the materials checked. Anthropic, OpenAI, Google: disclosed AI subprocessors, not three identified models.
- Recordings and text
- Recordings: Cloud, infrastructure in the US. Transcripts: Cloud, infrastructure in the US.
- Processing and training
- Fathom cloud service; STT engine not established. Cloud AI; the subprocessor list includes Anthropic, OpenAI and Google. In-house training on data without identifiers, with opt-out
- Reference plan
- Premium · annual billing: $192.00 in the first year; $192.00 in the following year at the same prices. The Team plan requires at least 2 seats; we do not present the price of one seat as the cost of an entire team.
- Average annual cost over 2 years
- $192.00
- Features for the price
- Premium for one person. Personal voice profile on iOS, translated summaries and limited live features on Mac; no Team/Business features.
- Free option
- Free · ongoing plan with no minute allowance. Recordings, transcripts, basic conversation summaries, clips, search and text copying. 5 advanced summaries per month, then the basic template; the full AI feature set is in Premium. This is not an importer for an arbitrary audio archive. Applies to recording supported meetings. Transcripts can be copied to the clipboard; we do not promise transcript file downloads.
- Meetings
- Assistant with recording; also a described bot-free mode on Mac (beta).
- Files and export
- No external file import according to the help documentation. Export and integrations do not remove this limitation.
- Teamwork
- Shared features and administration in Team/Business. The Premium price shown is for one person.
- Mobile models
- Remote and local models are treated equally. Model selectors and specific versions depend on the client and plan.
- Automation
- Webhook when a meeting is ready; integrations such as n8n.
Category scores for Fathom| Category | Score |
|---|
| Availability | 4 / 5 |
|---|
| Internationalization | 2.2 / 5 |
|---|
| AI models | 3–5 / 5 |
|---|
| Functionality | 4 / 5 |
|---|
| Live features | 2–4 / 5 |
|---|
| Privacy / security | 1 / 5 |
|---|
| Cost | 2 / 5 |
|---|
| Files and export | 3–4 / 5 |
|---|
| Meetings and collaboration | 5 / 5 |
|---|
| Agents and automation | 5 / 5 |
|---|
| Free mode | 5 / 5 |
|---|
| Features for the price | 1.9–2.2 / 5 |
|---|
Manufacturer sources

Notta
4.1 / 5 · 12/12 categories
App website · App Store
Baseline average, without priorities, using a scenario exchange rate of 1.10 USD/EUR. Average range: 4–4.3. Interactive values may change when weights are selected.
- English transcription
- Yes · Cloud mode; local mode has a separate list · Source
- Who it suits
- Meetings, imports and translation, with a separate local desktop mode.
- Main limitation
- Privacy Mode has fewer languages and features. Live translation may require an add-on.
- Availability
- Web, mobile and desktop. Private mode is described for desktop only.
- Transcription languages
- 58. Polish in the cloud. Local mode: 4 languages with the proprietary engine or 11 with Apple Speech Analyzer, excluding Polish. Translation has separate lists.
- Interface
- 21 localizations · iOS · Notta 6.81.24. Localization list declared by the vendor in the App Store, checked on October 10, 2026. This is not an audit of translation completeness. 21 localizations, including two Chinese script variants. The same list is not assumed for web / desktop.
- Features
- Some AI and translation features require add-ons or have separate limits. They are not equally available in local mode.
- Live
- Transcription and translation can work live. According to the instructions and integration FAQ, AI Notes require a completed transcript, so they are not counted as live summaries. Automatic generation after a meeting does not mean updates during the conversation.
- Cloud models
- Notta cloud STT: 58 service languages; model name not established. Cloud AI / Notta Brain; current model versions not established.
- Recordings and text
- Recordings: Cloud (help documentation indicates Tokyo) or local Privacy Mode. Transcripts: Cloud or local in Privacy Mode.
- Processing and training
- Cloud or local Privacy Mode on desktop. AI features in the cloud; Privacy Mode has limited scope and is not a full local equivalent. Full cloud training terms not established in this research
- Reference plan
- Pro · annual billing: $97.99 in the first year; $97.99 in the following year at the same prices. Translation add-ons at USD 72/108 per year are not included in the base USD 97.99.
- Average annual cost over 2 years
- $97.99
- Features for the price
- Pro. Translation of completed transcripts is included; live translation is excluded as a paid add-on. The separately paid Notta Brain is not counted as confirmed chat in Pro.
- Free option
- Free · 120 min/month, up to 3 min per conversation. 50 file imports and 10 AI summaries per month; online meeting transcription. Copy part or all of the available text on web and mobile, including on Free. 3 min of transcription per conversation: does not meet the full 30 min threshold. Copying is free; downloading a transcript as a file is a separate paid feature. The minute allowance, number of imports and length limit are separate restrictions.
- Meetings
- Notta Bot or bot-free desktop (beta).
- Files and export
- Audio/video import; DOCX, TXT, SRT, PDF, XLSX. Limits depend on the plan.
- Teamwork
- Collaboration and integrations; team management and SSO in the relevant plans.
- Mobile models
- Remote and local models are treated equally. Model selectors and specific versions depend on the client and plan.
- Automation
- Zapier and Notta automation.
Category scores for Notta| Category | Score |
|---|
| Availability | 5 / 5 |
|---|
| Internationalization | 4.4 / 5 |
|---|
| AI models | 3.5–5 / 5 |
|---|
| Functionality | 4–5 / 5 |
|---|
| Live features | 4 / 5 |
|---|
| Privacy / security | 3 / 5 |
|---|
| Cost | 3 / 5 |
|---|
| Files and export | 5 / 5 |
|---|
| Meetings and collaboration | 5 / 5 |
|---|
| Agents and automation | 5 / 5 |
|---|
| Free mode | 4 / 5 |
|---|
| Features for the price | 2.2–2.6 / 5 |
|---|
Manufacturer sources

Sonix
3.0 / 5 · 12/12 categories
App website
Baseline average, without priorities, using a scenario exchange rate of 1.10 USD/EUR. Average range: 2.7–3.5. Interactive values may change when weights are selected.
- English transcription
- Yes · Source
- English interface
- Not confirmed · Source
- Who it suits
- Editorial work with files, captions and translations.
- Main limitation
- Cost depends on volume, and translation uses the hour allowance.
- Availability
- The availability rating concerns native apps, not whether a website can be opened on a given operating system.
- Transcription languages
- 54+. The developer’s claim for transcription. Interface and translation need separate checks.
- Interface
- Not established localizations · Web. An English login screen is confirmed; no complete official list of editor languages was found. 54+ refers to speech, and localized marketing pages do not establish app localization.
- Features
- AI analysis, transcription, captions and translation. Voice profiles across recordings have not been confirmed.
- Live
- The vendor states that live captions/transcription and translation for international audiences are available through limited Enterprise access (early access, on request). Both features therefore receive partial status. They are not generally available in the reference plan. Live summaries remain unconfirmed; the early access program was not tested.
- Cloud models
- Sonix STT: 54+ service languages; model identifier not established. AI Analysis: cloud; models and count not established. Articles comparing LLMs do not prove their use in the product.
- Recordings and text
- Recordings: Cloud, AWS US. Transcripts: Cloud, AWS US.
- Processing and training
- STT in the Sonix cloud service. Cloud translation and AI analysis. The developer states that it does not train on user content
- Reference plan
- Core · 60 hours per year: $275.00 in the first year; $275.00 in the following year at the same prices. This is an annual allowance, not 60 hours a month. PAYG may be cheaper for low volumes.
- Average annual cost over 2 years
- $275.00
- Features for the price
- Core, 60 hours per year. Translation uses the same allowance. Live transcription and translation in Enterprise early access are not included at this price.
- Free option
- One-time trial · 30 min. 30 minutes of trial transcription; no payment card required. The trial allowance does not renew each month. Further processing is paid. PAYG is an alternative to a subscription, but it is not a free plan.
- Meetings
- Mainly importing existing recordings; we do not equate this with a recording bot. Separately: the vendor states that Zoom/Meet/Teams integrations for live features are available in the Enterprise early access program, not in the standard plan.
- Files and export
- Editor, captions, translation and automation. Check export/integration scope for the plan.
- Teamwork
- Editorial collaboration and roles; advanced settings may require a higher tier.
- Mobile models
- Sonix Recorder works in a phone browser and as a PWA on iOS/Android. After recording stops, it uploads the file for cloud transcription. This confirms mobile STT access, not a native app, a mobile LLM or engine selection.
- Automation
- Webhooks, Zapier and automatic recording import.
Category scores for Sonix| Category | Score |
|---|
| Availability | 1 / 5 |
|---|
| Internationalization | 2.8–4.4 / 5 |
|---|
| AI models | 2.5–5 / 5 |
|---|
| Functionality | 4–5 / 5 |
|---|
| Live features | 2–4 / 5 |
|---|
| Privacy / security | 2 / 5 |
|---|
| Cost | 1 / 5 |
|---|
| Files and export | 5 / 5 |
|---|
| Meetings and collaboration | 5 / 5 |
|---|
| Agents and automation | 5 / 5 |
|---|
| Free mode | 1 / 5 |
|---|
| Features for the price | 1.6–1.9 / 5 |
|---|
Manufacturer sources

Descript
3.6 / 5 · 11/12 categories
App website
Baseline average, without priorities, using a scenario exchange rate of 1.10 USD/EUR. Average range: 3.3–3.7. Interactive values may change when weights are selected.
- English transcription
- Yes · Source
- English interface
- Not confirmed · Source
- Who it suits
- Text-based editing, audio enhancement and media production.
- Main limitation
- A broader editor, not just a notebook. Media and AI have separate limits.
- Availability
- Mac / Windows and web on a supported computer. Mobile browsers and tablets are not officially supported for editing and account features; playback alone may work.
- Transcription languages
- 26 languages · 27 variants listed. Current help: 26 languages, including Polish. The list includes two English variants (27 entries). Greek, Hindi and Welsh are marked as beta. One language per file; some editing tools are English-only. This is not the dubbing list.
- Interface
- Not established localizations · Desktop / web. A UI localization request remains in the official feedback system, but it is not a complete, current list. We do not turn an “English only” assumption into a confirmed count.
- Features
- Text-based editing and generative features. Voice cloning is not the same as recognizing a speaker.
- Live
- Recording and editing do not prove that transcription or translation is displayed during the conversation.
- Cloud models
- Engine selection in App Settings > AI models > Transcription. Multiple options are confirmed, but the public instructions do not provide a full list of names and versions. Underlord offers a choice of models. Help documentation includes Claude Haiku 4.5 as an economical option; the full current catalog needs confirmation.
- Recordings and text
- Recordings: Cloud service; region not established in this research. Transcripts: Cloud service with a desktop client.
- Processing and training
- STT and audio tools in the cloud service. Cloud Underlord and AI tools; a desktop app does not imply a local LLM. Content used for research and development only with consent to data sharing (opt-in); this does not cover training production AI models.
- Reference plan
- Hobbyist · annual billing: $192.00 in the first year; $192.00 in the following year at the same prices. The price also includes the editor and creative tools; it is not the cost of transcription alone.
- Average annual cost over 2 years
- $192.00
- Features for the price
- Hobbyist: 10 media hours and 400 AI credits per month. Video editing and generative tools do not receive additional points here.
- Free option
- Free · 60 media minutes/month. Recording and text-based audio/video editing; limited access to AI tools. 60 minutes of media per month; restrictions on AI and video export. Free entitlement to text export has not been separately confirmed. Media minutes are not a separate additional transcription allowance. Free video exports may have a watermark.
- Meetings
- Recording in Descript / Rooms and media import; not automatically a meeting bot.
- Files and export
- Media import and export, text-based editing and captions. Strong editing focus.
- Teamwork
- Collaboration and permissions depend on the plan. An entire team is not priced at the cost of one Hobbyist license.
- Mobile models
- The vendor does not support editing or account features in mobile browsers or on tablets. Possible playback access is not a transcription client.
- Automation
- Official Zapier integration: import, editing, publishing and export actions. The trigger comes from another app; Descript does not provide its own Zapier triggers.
Category scores for Descript| Category | Score |
|---|
| Availability | 3 / 5 |
|---|
| Internationalization | 2.2–3.8 / 5 |
|---|
| AI models | 3 / 5 |
|---|
| Functionality | 4–5 / 5 |
|---|
| Live features | N/A |
|---|
| Privacy / security | 2 / 5 |
|---|
| Cost | 2 / 5 |
|---|
| Files and export | 5 / 5 |
|---|
| Meetings and collaboration | 5 / 5 |
|---|
| Agents and automation | 5 / 5 |
|---|
| Free mode | 4–5 / 5 |
|---|
| Features for the price | 1.8–2.4 / 5 |
|---|
Manufacturer sources

Happy Scribe
3.7 / 5 · 12/12 categories
App website · App Store
Baseline average, without priorities, using a scenario exchange rate of 1.10 USD/EUR. Average range: 3.4–4. Interactive values may change when weights are selected.
- English transcription
- Yes · Source
- Who it suits
- Captions, content localization and the option of human proofreading.
- Main limitation
- Separate import and meeting limits; human services cost extra.
- Availability
- iOS and Android confirmed in the changelog. Desktop: the changelog describes features, while pricing still says Soon; Mac / Windows status remains unestablished.
- Transcription languages
- 150+ languages and dialects. 150+ languages / dialects in the offering. The changelog describes 99 meeting languages; the product page mixes 99+ and 150+. We do not derive a single model’s language count from this.
- Interface
- 1 localizations · iOS · HappyScribe: AI Notetaker. Localization list declared by the vendor in the App Store, checked on October 10, 2026. This is not an audit of translation completeness. The complete UI language list for the web editor and Android is unconfirmed. The 150+ languages / dialects advertised are not menu languages.
- Features
- Transcription, captions, translation and AI. Human proofreading is a separate service.
- Live
- The vendor states that live meeting transcription is unavailable and that notes and summaries are delivered after the conversation. This finding is not applied to live translation capabilities, which the documentation does not settle.
- Cloud models
- Multiple models for EN / FR / ES / DE / DA; count and names undisclosed. Automatic selection, not a confirmed user choice. Ask AI: 4 added models: Claude Opus 4.8, GPT-5.5, GLM 5.2, Mistral Large. An update list is not necessarily the full catalog.
- Recordings and text
- Recordings: Claimed EU storage. Transcripts: EU; possible subprocessors outside the EEA.
- Processing and training
- Cloud STT; combining results from several models is described for 5 languages. Cloud Ask AI, summaries and translation. Opt-in training, potentially also with partners
- Reference plan
- Basic · annual billing: $102.00 in the first year; $102.00 in the following year at the same prices. Overages USD 0.20/min; human transcription from USD 2/min, outside the AI subscription.
- Average annual cost over 2 years
- $102.00
- Features for the price
- Basic: 24 hours of imports per year and analysis of 10 files per month. Human proofreading and overages are outside the price.
- Free option
- Free · meetings up to 45 min + 10 min import trial. Ongoing free meeting transcription, speaker labels and post-meeting notes. Unlimited meetings of up to 45 min each; TXT and SRT in the Free plan table. Limited history and AI tools. Transcription, subtitles and translation of imported files: a separate 10-minute trial, not a free monthly allowance. Points for 30-minute content and more than 60 min/month apply to meetings, not arbitrary files. The official Free description confirms transcription, not just audio recording. The score covers the basic text workflow.
- Meetings
- Notetaker joins as a bot; audio or audio/video. Also mobile recordings.
- Files and export
- Captions, editorial formats, translation and collaboration. Import minutes are not the same as meeting time.
- Teamwork
- Collaboration and roles in team plans. Basic: 1 seat; Pro: 3, Business: 5.
- Mobile models
- Remote and local models are treated equally. Model selectors and specific versions depend on the client and plan.
- Automation
- Signed partner webhooks when a transcript / summary is ready.
Category scores for Happy Scribe| Category | Score |
|---|
| Availability | 3–5 / 5 |
|---|
| Internationalization | 3.4 / 5 |
|---|
| AI models | 3.5–5 / 5 |
|---|
| Functionality | 4–5 / 5 |
|---|
| Live features | 1–2 / 5 |
|---|
| Privacy / security | 2 / 5 |
|---|
| Cost | 2 / 5 |
|---|
| Files and export | 5 / 5 |
|---|
| Meetings and collaboration | 5 / 5 |
|---|
| Agents and automation | 5 / 5 |
|---|
| Free mode | 5 / 5 |
|---|
| Features for the price | 2.2–2.6 / 5 |
|---|
Manufacturer sources

Plaud
3.6 / 5 · 12/12 categories
App website
Baseline average, without priorities, using a scenario exchange rate of 1.10 USD/EUR. Average range: 3.5–3.9. Interactive values may change when weights are selected.
- English transcription
- Yes · Source
- Who it suits
- A separate device for recording conversations and meetings.
- Main limitation
- Additional hardware purchase; cloud processing even without Cloud Sync.
- Availability
- Mobile apps, web and Plaud Desktop for Mac/Windows. The comparison covers the ecosystem with Note hardware.
- Transcription languages
- 112+. The developer’s transcription claim, not the number of interface languages.
- Interface
- 10 localizations · Plaud App (iOS / Android) and Desktop. 10 localizations from two official lists, including simplified and traditional Chinese. Polish is not listed. Each client has its own language setting; web requires separate confirmation.
- Features
- Auto speaker labeling: your own voice profile and learning labels for other speakers; Sync speaker labels carries identification across recordings. Vendor help rules out native translation between languages; retranscription is not that feature.
- Live
- The vendor rules out live transcription: transcripts and summaries are generated after recording ends. It also does not offer native translation. Offline recording on the device is neither offline transcription nor live AI.
- Cloud models
- 112 service languages; exact STT model and version not established. The AI offering lists GPT-5.5, Claude Sonnet 4.6 and Gemini 3.1 Pro. Three named models, with no claim of a best one.
- Recordings and text
- Recordings: Regional cloud; Frankfurt claimed for the EU. Transcripts: Cloud and app; depends on Cloud Sync.
- Processing and training
- Recording on the device; cloud transcription even with Cloud Sync off. Cloud models generating summaries and answers. In-house training after opt-in; LLM providers do not train
- Reference plan
- Note USD 159 + Pro USD 99.99/year: $258.99 in the first year; $99.99 in the following year at the same prices. The first year includes hardware; subsequent years include only the plan. No bundle promotions; phone/computer not included.
- Average annual cost over 2 years
- $179.49
- Features for the price
- Note + Pro, 1200 minutes per month. Hardware cost is included in year one; Team features are excluded.
- Free option
- Starter · 300 min/month after purchasing a device. Transcription and summaries, templates, export and sharing; no additional Starter subscription fee. 300 min/month. The maximum length of a single Starter transcript is unconfirmed. Requires ownership of a Plaud device. This is a free service for purchased hardware, not a free bundle.
- Meetings
- Mac / Windows desktop captures audio without a bot; requires a Plaud device linked to the account.
- Files and export
- Single and bulk export in the app, web and desktop: TXT/SRT/DOCX/PDF transcripts, summaries also in Markdown; audio in MP3 or WAV depending on the client. Zapier starts the next workflow after results are ready.
- Teamwork
- The separate Team plan has a shared workspace and seat, billing and device management. It is not covered by the reference Note + Pro price. Recordings remain private until shared.
- Mobile models
- Remote and local models are treated equally. Model selectors and specific versions depend on the client and plan.
- Automation
- Official Zapier sends completed transcripts and summaries.
Category scores for Plaud| Category | Score |
|---|
| Availability | 5 / 5 |
|---|
| Internationalization | 4.6 / 5 |
|---|
| AI models | 4–5 / 5 |
|---|
| Functionality | 4 / 5 |
|---|
| Live features | 1 / 5 |
|---|
| Privacy / security | 2 / 5 |
|---|
| Cost | 2 / 5 |
|---|
| Files and export | 5 / 5 |
|---|
| Meetings and collaboration | 5 / 5 |
|---|
| Agents and automation | 3–5 / 5 |
|---|
| Free mode | 4–5 / 5 |
|---|
| Features for the price | 1.9 / 5 |
|---|
Manufacturer sources

TurboScribe
3.2 / 5 · 11/12 categories
App website
Baseline average, without priorities, using a scenario exchange rate of 1.10 USD/EUR. Average range: 2.7–3.7. Interactive values may change when weights are selected.
- English transcription
- Yes · Source
- Who it suits
- Large audio / video archives and subtitle export in a simple web service.
- Main limitation
- Cloud processing instead of offline use; bot features or a full AI notebook are not assumed.
- Availability
- Browser-based service. No native apps in the version described.
- Transcription languages
- 98+. Transcription and translation have separate language lists. The vendor states that text can be translated into 134+ languages.
- Interface
- 41 localizations · Web · User Interface Language selector. 41 localizations counted in the UI selector, including 2 Chinese variants. Polish confirmed on the transcription and login forms. Not all screens behind login were audited.
- Features
- Speaker separation confirmed; persistent voice identification unknown. Features of an external GPT are not automatically counted as native.
- Live
- Transcription of uploaded files is confirmed; insufficient evidence to assess a complete live mode.
- Cloud models
- Whisper in the cloud; 3 speed / accuracy modes, versions undisclosed. LLM names and model count are unconfirmed. A link to TurboScribe GPT does not establish a built-in LLM.
- Recordings and text
- Recordings: Cloud, encrypted at rest. Transcripts: Cloud, encrypted at rest.
- Processing and training
- STT on machines owned or controlled by TurboScribe; no external transcription APIs according to the FAQ. Translation handled by the service; providers unknown. The vendor FAQ rules out model training on files and transcripts.
- Reference plan
- Unlimited · annual billing: $120.00 in the first year; $120.00 in the following year at the same prices. One person; USD 120 billed annually. This is not a human transcription service.
- Average annual cost over 2 years
- $120.00
- Features for the price
- Unlimited. External chat and linked services are not counted as confirmed AI included in the subscription.
- Free option
- Free · 3 files per day, up to 30 min each. File transcription, speaker separation and text / subtitle export. Lower processing priority; a limit of 30 min per file and 3 files per day. Longer files and larger batches require Unlimited. Renews daily. Meets the 60 min/month threshold in our scale, but is not unlimited.
- Meetings
- Import existing audio / video files.
- Files and export
- Import through the web interface; text and subtitle export. The vendor explicitly rules out an API and automated access to the service.
- Teamwork
- Separate Teams version: shared billing, adding and removing users. This confirms access administration, but does not establish collaborative transcript editing or CRM integration. The reference price covers one seat.
- Mobile models
- The vendor confirms all web features on phones: import, STT, translation and export. A home screen shortcut is not a native app and does not increase the availability score. Model selection and a built-in LLM remain unconfirmed.
- Automation
- No official file upload automation; users work through the web interface.
Category scores for TurboScribe| Category | Score |
|---|
| Availability | 1 / 5 |
|---|
| Internationalization | 4.4 / 5 |
|---|
| AI models | 2–5 / 5 |
|---|
| Functionality | 3–5 / 5 |
|---|
| Live features | N/A |
|---|
| Privacy / security | 2 / 5 |
|---|
| Cost | 2 / 5 |
|---|
| Files and export | 4 / 5 |
|---|
| Meetings and collaboration | 3–5 / 5 |
|---|
| Agents and automation | 2–3 / 5 |
|---|
| Free mode | 5 / 5 |
|---|
| Features for the price | 1.7–2.8 / 5 |
|---|
Manufacturer sources

Apple Notes
3.7 / 5 · 11/12 categories
App website · App Store
Baseline average, without priorities, using a scenario exchange rate of 1.10 USD/EUR. Average range: 3–4.2. Interactive values may change when weights are selected.
- English transcription
- Yes · Compatible hardware and device language · Source
- Who it suits
- Recordings and short summaries without purchasing a separate app.
- Main limitation
- Language and device restrictions; not a full transcription toolkit.
- Availability
- Also iPad. Transcription: iPhone 12+ and Mac M1+. Summaries require Apple Intelligence-compatible hardware.
- Transcription languages
- 9 languages / 10 language settings (2 Chinese variants). Notes list: EN, ES, PT, IT, FR, DE, JA, KO, simplified / traditional ZH. Apple Intelligence has a separate list.
- Interface
- 33 localizations · Apple Notes · iPhone / iPad. Localization list declared by the vendor in the App Store, checked on October 10, 2026. This is not an audit of translation completeness. 33 localizations, including two Chinese variants. A Polish interface does not imply Polish transcription.
- Features
- Recording transcription and summary. Not all Siri / Translate features are automatically attributed to Notes.
- Live
- Text during recording; the documented summary is generated after recording. Call translation in Phone is not attributed to Notes.
- Cloud models
- The Notes pages reviewed do not identify a specific model or the complete STT processing path. Apple Intelligence may use Private Cloud Compute; selected by the system, with no LLM selector.
- Recordings and text
- Recordings: Recording in a note; may be stored in iCloud. Transcripts: In the note / transcript, depending on account and sync settings.
- Processing and training
- Apple system STT; the exact Notes workflow was not established in this research. Apple Intelligence: on-device or Private Cloud Compute; we do not assume every summary is generated offline. Apple Intelligence: private content is not used for training unless the user consents.
- Reference plan
- Built in, on compatible hardware already owned: $0.00 in the first year; $0.00 in the following year at the same prices. USD 0 excludes the device purchase and any additional iCloud storage.
- Average annual cost over 2 years
- $0.00
- Features for the price
- On an existing compatible Apple device. Summaries require Apple Intelligence; languages and hardware restrict availability.
- Free option
- Built in · no separate app fee. Recordings, transcription and text copying in Notes on a compatible device. Summaries require Apple Intelligence. Supported languages and devices are limited. Numerical limits have not been established, so we do not automatically award points for their absence. The device cost and any iCloud charges are separate. Not every feature works on every iPhone/Mac.
- Meetings
- Recordings in Notes; Phone / FaceTime can save call recordings to Notes.
- Files and export
- After adding a transcript to the note body, it can be exported as Markdown on Mac (OS 26) or PDF. Markdown meets the structured format criterion, but is not a subtitle file with timestamps. Shortcuts operate on the note body; manually transferring the transcript means partial automation. Import and automatic STT of external audio remain unconfirmed.
- Teamwork
- Note sharing; this is not equated with meeting platform administration.
- Mobile models
- Availability depends on hardware, region and language.
- Automation
- The official Find Notes / Create Note / Append to Note Shortcuts actions can pass note content onward. A transcript must first be added to the note body: a partial workflow, not unattended access to the recording.
Category scores for Apple Notes| Category | Score |
|---|
| Availability | 3 / 5 |
|---|
| Internationalization | 2.6 / 5 |
|---|
| AI models | 3 / 5 |
|---|
| Functionality | 2–5 / 5 |
|---|
| Live features | 2–4 / 5 |
|---|
| Privacy / security | N/A |
|---|
| Cost | 5 / 5 |
|---|
| Files and export | 4–5 / 5 |
|---|
| Meetings and collaboration | 3–5 / 5 |
|---|
| Agents and automation | 3–5 / 5 |
|---|
| Free mode | 3–5 / 5 |
|---|
| Features for the price | 2.6–4.6 / 5 |
|---|
Manufacturer sources

Google Recorder
3.4 / 5 · 12/12 categories
App website
Baseline average, without priorities, using a scenario exchange rate of 1.10 USD/EUR. Average range: 2.8–4. Interactive values may change when weights are selected.
- English transcription
- Yes · Scope depends on the Pixel model · Source
- English interface
- Not confirmed · Source
- Who it suits
- Free conversation transcription on a compatible Pixel phone.
- Main limitation
- Not a general app for every Android device; Polish is absent from the lists reviewed.
- Availability
- Pixel 3+ / Pixel Tablet; summaries on Pixel 8+. Not every Android phone.
- Transcription languages
- 35 languages / 42 retranscription variants; 8 live languages on Pixel 6+. Counts taken from help center lists: retranscription may use the cloud; live support is narrower. Older Pixels support fewer languages.
- Interface
- Not established localizations · Google Recorder · Pixel. No complete official list of Recorder interface localizations was found. Android languages, translated help pages and retranscription languages are not substitutes; Polish UI is unconfirmed.
- Features
- Diarization on supported Pixels. The documentation states that voice models are deleted within a few minutes after labeling: this is not a persistent profile for automatic recognition across recordings. Summaries depend on the phone model and language.
- Live
- Live transcription; summaries available after saving. Live translation unconfirmed.
- Cloud models
- Retranscription may use Google servers. Summaries require the cloud for some languages; exact LLM version unknown.
- Recordings and text
- Recordings: On-device; optional Google account backup / sync. Transcripts: On-device or in the Google account backup.
- Processing and training
- Basic STT runs locally; retranscription may process audio on servers. Summaries depend on language / device; some run in an isolated cloud environment. Not all training policies for optional cloud features have been established.
- Reference plan
- Included with a Pixel already owned: $0.00 in the first year; $0.00 in the following year at the same prices. The purchase of a compatible device and optional cloud storage are excluded.
- Average annual cost over 2 years
- $0.00
- Features for the price
- On an existing compatible Pixel. Temporary speaker labels, with no persistent profile; summaries depend on the device and language.
- Free option
- Built in · no separate fee on Pixel. Recording, transcription and text / Google Docs export. Recordings up to 18 hours. Selected devices and languages; summaries have separate requirements. A monthly service allowance is not specified in the reviewed materials. Requires a compatible Pixel; the phone itself and backup storage are not free. Some features may use the cloud.
- Meetings
- A voice recorder on Pixel, not a meeting bot.
- Files and export
- Share audio, text files and Google Docs. Retranscribing a recording made in the app does not establish support for arbitrary audio imports.
- Teamwork
- Sharing through the account and Google Docs; not a full team platform.
- Mobile models
- The help center does not provide a complete catalog of LLM versions.
- Automation
- Export to Google Docs; automation triggers unknown.
Category scores for Google Recorder| Category | Score |
|---|
| Availability | 2 / 5 |
|---|
| Internationalization | 2.2–3.8 / 5 |
|---|
| AI models | 2–4 / 5 |
|---|
| Functionality | 3–4 / 5 |
|---|
| Live features | 2–4 / 5 |
|---|
| Privacy / security | 3 / 5 |
|---|
| Cost | 5 / 5 |
|---|
| Files and export | 2–5 / 5 |
|---|
| Meetings and collaboration | 3–4 / 5 |
|---|
| Agents and automation | 2–5 / 5 |
|---|
| Free mode | 4–5 / 5 |
|---|
| Features for the price | 2.8–4 / 5 |
|---|
Manufacturer sources
How should you read the scores and average?
By default, the arithmetic mean of all categories with data, equally weighted. Optionally, up to 3 priorities receive a weight of 1.5, with the rest weighted at 1. Result = sum(rating × weight) / sum of weights of rated categories. We use the midpoint of a range; this is a convention, not an estimate of likely quality. N/A is omitted along with its weight, never replaced by zero. We show completeness and the average range from the lower / upper bounds; it covers only rated categories. Categories overlap in part. This is neither a benchmark nor a comparable ranking when data completeness differs.
Availability
1 base point + 1 for each of four platforms with an app: Android, iPhone, Mac, Windows. Web is shown separately, without points. We do not assess feature parity between platforms.
Internationalization
Rating = 60% of transcription points + 40% of UI points, without rounding to whole stars. STT: 1–9 languages = 1★, 10–24 = 2★, 25–49 = 3★, 50–99 = 4★, 100+ = 5★. UI: 1 localization = 1★, 2–4 = 2★, 5–9 = 3★, 10–19 = 4★, 20+ = 5★. UI is counted for the specified client; writing-system variants count as separate localizations. This does not guarantee the same UI on all platforms. A missing component yields a 1–5 range, not zero; both unknown = N/A. We do not count the base model’s potential a second time. This measures coverage, not language quality; counts including dialects are marked.
AI models
The mean of phone and computer scores, weighted 50% each. An unknown platform contributes a 1–5 interval; both unknown means N/A. Each platform uses the same scale: Four capabilities: STT access, LLM access, a choice of at least 2 STT models / engines, a choice of at least 2 LLMs. Yes = 1; rating = 1 + round(sum). Local and remote models count equally. Unestablished choices produce a range, not zero. N/A when client data is missing. Automatic provider routing is not user choice. We do not rate model quality.
Functionality
Six equally weighted features: transcription, diarization, persistent voice recognition, summaries, AI questions, translation. Yes = 1, partial scope = 0.5, no = 0. Rating: 1 + round(4 × sum / 6). Unestablished fields produce a range, not zero.
Live features
Three features: live transcription, evolving live summary, live translation. Yes = 1, partial = 0.5, no = 0. Rating: 1 + round(4 × sum / 3). Missing data yields a range or N/A, not a negative rating.
Privacy / security
5★: available features processed locally by default, with no content received by the developer. 4★: local STT and AI can be configured, with optional cloud use. 3★: local mode with a limited scope. 2★: cloud, training disabled or opt-in. 1★: cloud with disclosed training / model improvement and opt-out. Optional backup is described separately. The score assesses content exposure, not resistance to attacks or GDPR compliance.
Cost
The cost used for scoring is C = (year 1 + year 2) / 2, in USD. 5★: C = 0; 4★: 0 < C ≤ 50; 3★: 50 < C ≤ 100; 2★: 100 < C ≤ 200; 1★: C > 200. A two-year period distinguishes a one-time purchase from a subscription. We assume prices remain unchanged. We score the specified plan, not its free alternative. EUR prices use an editable exchange rate (default: 1.10 USD/EUR, not a market rate). Plaud hardware is included; an existing phone/computer, country-specific taxes, overages and optional services are separate.
Files and export
Four capabilities: external audio import, text export, captions or structured formats, automation / API / CLI. Yes = 1, partial = 0.5. Rating: 1 + round(sum). Unestablished fields produce a range.
Meetings and collaboration
Four collaboration capabilities rated: shared workspace / editing, integrations with work tools, organization management, sharing or export. Yes = 1, partial = 0.5. Rating: 1 + round(sum). Recording method and Zoom / Meet / Teams are described as facts, without an extra point for each brand. Some features require a higher plan.
Agents and automation
Four channels: official MCP / agent connector, documented API or CLI, event-driven automation / Shortcuts, export for reuse. Yes = 1; rating = 1 + round(sum). Unestablished fields yield a range. We assess channel availability, not whether their capabilities are equivalent. File export does not establish MCP support. Read / write scope, plan and permissions are described separately.
Free mode
1 base point + 1 each for: ongoing use with no app/service fee, full transcription of a recording ≥30 min, a renewable allowance ≥60 min/month (or no allowance limit), and text export/copying. We count confirmed capabilities of the free version, not the paid plan. A time-limited trial or one-time allowance alone = 1★, regardless of trial features. Unconfirmed conditions produce a range, not zero. We assess use on existing compatible hardware; buying a Mac, Pixel or Plaud device is not free. AI, retention and formats are described separately, without extra points. This is an explicit scenario for basic work with text, not a rating of the plan’s overall value. Free mode and Annual price are separate categories: capabilities at zero cost and the cost of the specified paid offer.
Features for the price
A basket of 10 equally weighted features: 6 core features, 3 live features and text export/copying. Yes = 1, partial = 0.5, unavailable or a paid add-on outside the price = 0. F = sum / 10. Score = 1 + 4 × F / (1 + C / 100), where C is the average annual cost over 2 years in USD. A continuous 1–5 scale, with no rounding in the average; USD 100 is an explicit editorial reference point, not a market standard. We do not divide by the price, so a free tool does not automatically receive 5★. Unconfirmed fields produce lower and upper bounds; all unknown = N/A. We count features in the specified plan, not the vendor’s entire product range. Minute/credit limits still apply. This measures the breadth of features for the price, not transcription quality or value for every user. Video editing, integrations, privacy and platforms do not add points here; they have separate descriptions/categories. This category deliberately combines aspects also scored separately under Cost and Features.