Spokenwork

Getting started

Create an account with your email and a password, then open the confirmation link we send. Your CA$5.00 starter credit is added at that moment.
On New job, choose a file or drop it in. Anything up to three hours; audio or video.
Tick the pieces of writing you want. The estimate on the right shows the most you will pay.
If the recording has unusual names or terms, type them in the "names and terms" box. The transcription model uses them as a hint and spells them correctly far more often.
Click Transcribe and write. A 60-minute file usually takes under two minutes.
Open the job to copy any piece, or download the transcript as text or subtitles, each piece as Markdown, or everything as a zip.

Supported files

MP3, M4A, WAV, FLAC, OGG, AAC, WMA, MP4, MOV, MKV, WEBM and most other common formats. Video is fine; we extract the audio track. Up to 500 MB and three hours per file. Large files upload in small parts, so a slow or patchy connection only re-sends the part that failed, and pressing the button again continues where it stopped. To save time on a slow connection, upload the audio only: an MP3 at 64 kbps is about 0.5 MB per minute and speech still sounds fine.

Getting better results

Clean audio matters most. A phone recording in a quiet room beats a video call with echo.
Use the names-and-terms box. Product names, people's names, acronyms and jargon are where transcription makes mistakes. Listing them fixes most of it.
Set the language if you know it. Auto-detect is good, but a 20-second English intro on a French episode can confuse it.
Chapters and pull quotes rely on timestamps. They are only as accurate as the transcript, so give them a quick read before posting.

Frequently asked questions

Do you keep my recording?: No. The uploaded file is deleted as soon as the transcript exists, usually within a couple of minutes. The transcript and the writing are kept for 30 days so you can download them, then deleted.
Is my content used to train anything?: We do not train models on your recordings, transcripts or results, and we do not sell or share them for anyone else to train on. Your transcript goes to the AI providers only to produce your result, and their published API terms say they do not train on it either. We can tell you what those terms say and link to them, but we cannot audit the providers, so we do not promise more than that. The current terms and what each provider receives are set out on the Security page.
Why was I charged less than the estimate?: The estimate is a maximum. If a piece of writing fails to generate you are not charged for it, and the minute count is confirmed after we read the file.
A job failed. Was I charged?: No. Failed jobs are never charged. Try again; if it keeps failing, tell us the file type and length.
Can I get the transcript with speaker names?: Not yet in "Included" mode. In "your own keys" mode, adding a Deepgram key gives you speaker labels.
What is the difference between Standard and Premium?: Both use the same per-platform briefs and the same checks (verbatim quotes, real timestamps, platform limits, no AI clichés). Standard is written by the OpenAI family of models; Premium by the Anthropic family, which in our tests stays closer to the speaker's own voice and specifics. You choose per job; the per-piece price is the only difference.
What does "checked" mean?: After each piece is written we verify it mechanically: every pull quote must appear word for word in the transcript, every chapter timestamp must exist in the recording, X posts must fit the character limit, plain-text pieces must contain no Markdown, and a list of AI clichés must be absent. A piece that fails is rewritten once with the problems listed; if it still fails, the provable errors are removed and, if nothing usable remains, the piece is not delivered and not charged.
What is "your own API keys" mode?: You bring keys from Groq, OpenAI, Deepgram or Anthropic and pay them directly; we charge a smaller platform fee. Keys are encrypted at rest and never shown again. Switch modes in Account at any time.
Does credit expire?: No.
Can I get a refund?: Unused credit is refundable within 30 days of purchase, less payment-processing fees. Ask us.
Can I use this for patient or client recordings?: Please don't upload recordings that identify patients or clients. Spokenwork is built for content you intend to publish, not for records that fall under health-privacy rules.
Which languages work?: Transcription supports most major languages. The writing is produced in the language of the recording.
Is there an API?: Yes. Generate an access key in Account and call the same endpoints the app uses. Documentation is in progress; ask and we will send what you need.

Still stuck?

Signed in: use Support and mention the job. Otherwise contact us. A person replies, usually within one business day.