SonaFactEvidence from audio
SonaFactEvidence from audio

Transcribe Chinese audio to text

Upload a recording in Chinese and get a transcript with speakers told apart and a timestamp on every word. Check it in the editor while the audio plays, then export it.

Upload your file

The recording language is already set to Chinese — you can change it before you start. The file stays in your browser until you create a free account.

Drop audio or video here
MP3 · WAV · M4A · MP4 · MKV · OGG · FLAC · WMA · OPUS · OGA · AMR · 3GP · WEBM · MOV · AAC · M4V
Upload only recordings you have the right to use. Record other people only with their consent.
45 min ≈ 4 min

Three transcripts a day on the free plan, up to 30 minutes each.

What you get

  • Word-level timestamps, so subtitle splitting lands exactly on the word.
  • Speakers separated automatically, with names you can set once for the whole transcript.
  • SRT and VTT subtitles in the language of the recording, on every plan.
  • AI summaries and other documents can be written in Chinese or in any other of the hundred languages, whatever the language of the recording.

A real result from SonaFact

On 2 October 2026 we ran the first 75 s of a public recording in Chinese through SonaFact. This is the beginning of the result exactly as it came out, with nothing corrected. The recording is one person reading a Wikipedia article, so the transcript has one speaker.

0:00 · Speaker 1

这报题,出自中文维基百科自由的百科传书,此录音在2016年6月20日录制。北部 …

0:20 · Speaker 1

提供 …

0:36 · Speaker 1

俄国,是 …

0:47 · Speaker 1

一 …

0:52 · Speaker 1

个位 …

0:55 · Speaker 1

于非洲东北部亚丁湾西岸的国家,北面与阿拉图尼亚为邻,西部、西南及南部与阿塞俄比亚被连, …

Words: 103 · length: 75 s · marked uncertain by the engine: 8.7%. Uncertain words are underlined — in the editor you check them by playing the moment they were said.

Recording: Djibouti - Chinese spoken article (1 out of 3) — Carrotkit, CC BY-SA 4.0, Wikimedia Commons. This transcript excerpt is shared under the same licence.

Three steps from recording to finished work

  1. Upload anything recorded. Audio or video — up to 30 minutes per file on Free, 3 hours on Starter, 6 on Pro and 10 on Studio. The language and the number of speakers are detected for you, or you set them yourself when you already know. You can also record straight in the browser.
  2. Review in the editor. Play it back with the line highlighted, fix a word, rename a speaker everywhere, move a line to the right person.
  3. Turn it into finished work. Export DOCX or PDF, generate minutes, action items, subtitles, chapters or a call report.

Pricing →

Other languages

SonaFact recognises a hundred languages and detects the language of a recording automatically.