Transcription

Every word, on the clock.

Upload a call, a voice note, an interview or a meeting and get the text back with a timestamp on every word, in the language it was spoken.

Customer video call

isiZulu · detected

0:00

Caller Thandi M.

Sawubona. Ngicela usizo nge-akhawunti yami.

Hello. Please help me with my account.

0:04

Agent

Yebo, ngingakusiza. Ngubani igama lakho?

Yes, I can help you. What is your name?

Click any word to play from itSRT · VTT · TXT · JSON

What you get

Export SRT or VTT captions, plain text, or JSON with every word's timing for your own systems.

Click any word to jump to that moment in the audio. The transcript follows along as it plays.

Handles narrow-band telephone recordings as well as clean studio sound, and evens out quiet recordings so you can hear them.

Up to an hour per file. Progress shows part by part, and you can start listening before it finishes.

Words the system is less sure of are underlined, so reviewers know exactly where to look.

Not sure which language is on the recording? Leave it on automatic.

1Drop in your recordings — several at once

2Pick a language or let it detect one

3Read, listen and export

Review every customer call in writing, in the language it was spoken.

Captions for every video you publish, in the language it was spoken.

Turn field interviews into timed text you can quote and analyse.

Keep a written record of public hotlines and meetings in local languages.

Languages: isiZulu, isiXhosa, Sepedi, Sesotho, Setswana, isiNdebele (South Africa), isiNdebele (Zimbabwe), Kiswahili, Luganda, Lingála.

Which audio formats work?: MP3, WAV, M4A, AAC, OGG/Opus, FLAC, WebM, AMR and 3GP — up to 200 MB and 60 minutes per file.
How is it priced?: Priced per minute of audio and billed by the second, from prepaid credits in your dashboard. No subscription.
How long do you keep my audio?: Recordings are kept for 30 days so you can replay them, and you can delete any recording at any time.
Can it tell closely related languages apart?: Usually. For close relatives — Setswana and Sesotho, or Zimbabwean isiNdebele and isiZulu — choose the language yourself for the best transcript. isiNdebele (South Africa) is always chosen by hand.