Chinese recordings to English speech

A live conversation or a recorded file?

Choose a live interpretation app for a conversation happening now. For a saved Chinese voice message, interview or lesson, upload the recording here to create an English voice track. You can listen to a free preview of up to 30 seconds before generating the full file. This page covers the recorded-file workflow.

Mandarin, dialects and names

“Chinese” is a broad label for a recording. The recognition provider lists Mandarin Chinese and Cantonese separately, so identify which variety your speaker actually uses before judging the result. This page focuses on Mandarin; it does not promise coverage or equal quality for every Chinese dialect. A short preview is a useful first check, especially for regional speech or a clip that switches languages.

For an interview or lesson, check names and specialist terms against the original recording. Decide how personal names, company names and place names should appear in English, and use that spelling consistently when editing the translated sentences. Do not infer a person's preferred English name from an automatically generated transcript.

Mandarin text does not use spaces between words in the same way English does. Listen across the full sentence when reviewing a subtitle break rather than treating each visible character as a separate word. For a product demonstration, check that the English instruction still refers to the right button, model or part. An English voice track should convey the instruction, without inventing details the speaker did not supply.

Language background: Scribe language and spacing documentation.

How to translate the recording

1. Upload a file you have permission to use. MP3, WAV and M4A audio, up to 200 MiB and 10 minutes.
2. Try the English preview. English is preselected. The source language is detected automatically; the preview covers up to 30 seconds and uses no credits.
3. Generate and review the full dub. Compare the source and English sentences, then edit translated names, terms or phrasing where needed. Only edited sentences are regenerated.
4. Download the result. Get English audio mixed with the background, voice-only audio and separate source/English SRT and VTT files. Audio uploads do not create video.

Voices, background and timing

Stock voices are assigned by detected speaker, with reuse if the voice catalogue is exhausted. Optional cloning requires your consent and a single identified speaker with at least 10 seconds of usable sample speech. Multiple speakers or insufficient sample speech fall back to stock voices. Review the preview and the full recording before sharing.

The translated speech is mixed with the separated background. Separation can leave artifacts, and timing adaptation cannot guarantee exact alignment for short or overlapping lines. Listen to sentence transitions and check that the English remains understandable at the resulting pace.

Related recordings, pricing and permissions

The audio translator explains the general file workflow. The Spanish recording guide covers another source language; the related guides below cover the other audio pairs.

Check the credit quote before full generation; charges depend on source duration and target languages. See pricing. Upload only recordings you own or have rights to use, including the voices, music and images. AI transcription, translation and pronunciation can contain mistakes, so review the result before publishing. See the terms. This tool processes recordings, not live calls.