Turn audio and video recordings into editable text, choose a transcription model, and export documents or timed subtitles. Typist also offers browser recording and read-only access to your transcript library from AI clients.
Used this tool? Rate it
Used this tool? Rate it
Typist is an AI audio and video transcription service operated by Kollektiv LLC. It turns a recording into editable text that can be searched, quoted and repurposed. The product combines file transcription with a browser recorder, a transcript library and a connection that lets compatible AI clients read completed transcripts.
Audio transcription is its central task, with document exports for interview and lecture work and timed text for video captions.
Typist offers xAI Grok, ElevenLabs Scribe and OpenAI Whisper, with language-specific labels for Most accurate, Fastest and Lowest cost. Those labels describe the vendor's intended tradeoffs for a particular language, rather than an independently verified ranking of every recording or accent.
Typist accepts MP3, M4A, WAV, FLAC, OGG and video files through its upload control. The published audio-to-text flow is:
Typist also provides a microphone recorder with a Start Recording control; transcription begins after the user stops recording.
Typist positions interview transcription for research workflows, qualitative coding and investigative journalism, with timestamped text that can be cited, coded or quoted.
Every interview sentence is described as carrying a timestamp for citations and quotes. This makes the time reference part of the published interview workflow.
Typist positions lecture transcription around recordings made in class, Zoom recordings and M4A files, producing searchable, timestamped text. The use case is access to what was said in a recording, with a written form that can be read again and located by time.
The free YouTube transcript utility retrieves a video's captions or closed captions and supports copying with or without timestamps, TXT/SRT/VTT downloads, text search, seeking to a moment and sharing a public page. That caption-retrieval route is different from uploading a private audio file for a selected speech-recognition model.
The YouTube summary utility produces a TL;DR, key topics, key moments with clickable timestamps, and structured markdown for AI clients.
The voice memo workflow uses a browser on a phone or laptop and starts by sharing the original recording as a file.
Typist MCP uses Streamable HTTP and browser-based OAuth, without requiring the user to create or copy an API key. A compatible AI client connects to the remote service, and the user signs in through the browser to authorize access. The public AI entry mentions clients such as ChatGPT and Claude; actual access depends on a client supporting the required remote transport and authorization flow.
The public monthly cards showed the following prices and limits as captured on October 8, 2026. Paid cards explicitly inherit the preceding plan's features; the table preserves that inheritance where a card does not repeat a value.
| Plan | Monthly price | Included usage | File limit | File retention | Concurrent transcriptions |
|---|---|---|---|---|---|
| Free | $0 | 60 free minutes advertised; every transcription model | Up to 500 MB | 7 days | Not stated in the card |
| Lite | $6.99/month | 10 credits per month; inherits Free | Up to 5 GB | 30 days | 3 |
| Premium | $19.99/month | 75 credits per month; inherits Lite | Inherited from Lite | Unlimited | 10 |
| Max | $49.99/month | 200 credits per month; inherits Premium | Inherited from Premium | Inherited from Premium | Inherited from Premium |
The Free card lists TXT, DOCX, PDF, SRT and VTT exports. Premium adds unlimited AI Insights, described as TL;DR, chapters, quotes and actions. Max adds priority support.
One credit covers 1 hour on Whisper v3 Turbo, 30 minutes on Grok 2.0 or Whisper v3, or 15 minutes on Scribe v2; credits reset each billing cycle, and speaker labels and filler removal work with Grok 2.0 or Scribe v2. The purchased unit is consequently a credit with a model-dependent duration, rather than a uniform hour across all engines.
The free headline is specifically advertised as 60 minutes. Because the same pricing section explains different credit consumption for different models, that headline alone does not establish identical free recording time on every engine. Monthly credit resets also describe a billing-cycle allowance rather than a lifetime pool of purchased transcription time.
Monthly or yearly subscriptions automatically renew, can be cancelled in account settings, and remain active until the billing period ends after a mid-period cancellation. Cancelling therefore ends the next renewal rather than automatically ending current-period access or reversing a completed purchase.
The stated 7-day refund window covers transcription-preventing service or technical failures, billing errors or duplicate charges, and first-time subscribers unsatisfied with transcription quality. The window is conditional, rather than an unconditional refund promise for every subscription.
Refund exclusions include successfully completed transcriptions, usage beyond free limits, requests after 7 days, buyer's remorse and changed circumstances. The eligibility language and the exclusions appear together in the terms; the stated quality-related category does not erase the separate completed-transcription or usage exclusions.
A refund request requires contacting support within 7 days with order details and a reason; the terms state a response within 48 hours.
Fees exclude applicable taxes unless stated otherwise. Subscription price changes apply at renewal with reasonable notice of increases.
The same distinction between a capability and its conditions applies to model labels. The service provides choices with different feature sets and consumption rates, and a named model's language coverage is not an independently tested accuracy percentage. The reviewed public material does not supply a product-level benchmark that would turn the promotional accuracy language into a reliable result for every uploaded file.
Users retain ownership of uploaded content and own the transcriptions generated from it.
Uploading grants Typist a limited, non-exclusive license solely to provide transcription, ending when the content or account is deleted.
The audio-to-text entry claims files are uploaded only to transcribe, removed afterward, and never sold, shared or used to train models. This is vendor privacy marketing attached to that conversion entry. It is not an independently verified statement about every downstream provider or every route through the product.
The privacy policy separately states that uploaded audio and resulting transcripts are stored, associated with the account, and deletable by users. The account-storage statement and the plans' retention periods do not describe the same immediate-removal promise as the conversion entry. The public wording therefore leaves different storage descriptions in place rather than a single universal deletion timeline.
The privacy policy says database, authentication, transcription, hosting, monitoring, analytics and payment providers access information as needed for their functions. That service-provider disclosure also differs in scope from the conversion entry's unqualified no-sharing wording. The training statement is the vendor's declaration; the reviewed privacy text does not separately specify the training policies of each processing provider.
Account deletion leads to personal-data deletion within 30 days, except required legal or regulatory retention; aggregate anonymized analytics may remain. This account-deletion rule is separate from deletion of an individual file, the retention allowance of a plan and expiry of a temporary export link.
The privacy policy claims encryption in transit and at rest, secure authentication and regular security monitoring. It also says providers may store and process data in countries including the United States.
Most free-tool outputs are automatically deleted within 15 minutes to 1 hour, while YouTube transcripts may instead be stored and displayed publicly and are subject to removal requests. The short temporary-output range covers downloaded files, converted media and similar free-tool outputs as described in the terms; it does not replace the account library's file-retention rules.
Typist's published MCP tools are read-only. They search, read and provide transcript exports; they cannot create, rename, edit or delete transcripts, and the scope excludes recordings and billing.
The YouTube transcript utility supports public pages and links. Its terms allow transcripts from public videos to be stored and displayed for sharing and discovery, unlike the separate descriptions of account storage and plan retention.
The subtitle entry says only the video's audio track is transcribed and the picture is ignored. It describes word-level timing for subtitle cues, rather than visual analysis of scenes or video frames.
