Translate Audio Files to English

Read it in English or hear it in English

Most recordings that need English are meant to be understood once: a voicemail, an interview, a lecture, a relative's story. For those, English text is enough. A voice track is for recordings you want to publish to English listeners, such as a podcast feed, a training recording or a talk.

English text

The original-language transcript, the English translation and English SRT/VTT subtitles, at 1 credit per minute; leave "Audio and text output" switched off

English voice

Everything above plus an M4A with English speech over the original background sound, at 50 credits per minute; switch "Audio and text output" on

Both outputs come from the same upload, so you can start with the text, read it, and run the file again for the voice only if the recording turns out to be worth publishing.

Recordings that work straight off a phone

The format matters more than the app that made it. These are the usual sources and whether they can be uploaded as they are:

iPhone voicemail

Shared from the Phone app as an M4A file, which works

iPhone Voice Memos

M4A, which works

Google Voice voicemail

Downloaded as MP3, which works

Telegram voice message

Saved as OGG, which works

Zoom audio-only recording

M4A, which works

WhatsApp voice notes saved as .opus

Not accepted; convert to MP3 or M4A first

Android call recorders and voicemail apps (.amr, .3gp)

Not accepted; convert to MP3 or M4A first

Accepted audio formats are MP3, WAV, M4A, AAC, OGG, FLAC, WebM, up to 2 hours per file. A recording you took as a video goes to the video translator for English instead, which keeps the picture on your device and returns an English MP4. Phone recordings are often quiet or narrow-band; recording audio on a phone for transcription explains what helps.

How the source language affects the English

The English can only be as good as the transcript underneath it, and transcription quality varies by language and by recording. Widely spoken languages recorded clearly give the cleanest drafts. Three situations need a closer read:

• Mixed-language speech, such as Spanglish, Hinglish or Arabic mixed with French, produces more errors than speech in one language. The words are written as heard and then translated, so check the switches.
• Chinese means Mandarin. Cantonese recordings aren't a supported source and come out unreliable.
• Heavy dialect, fast overlapping talk and phone-quality audio all lower accuracy before translation even starts.

Language-specific notes live on the pair pages: Spanish audio to English, Hindi audio to English, and the video pages for Japanese, Korean, Chinese, Arabic, Russian, French and German, which apply to audio too.

Long recordings and other tools

Files can run to 2 hours, and you pay per minute from credits with no subscription. Other to-English tools set different limits; as checked on their own pages on 5 October 2026, TurboScribe's free tier allows 3 files a day of up to 30 minutes each and its paid plan accepts files up to 10 hours, and it also accepts .opus and .3gp files and labels speakers. Some free tools run Whisper entirely in your browser and never upload the audio, which suits short private clips if your computer is fast enough. Pick by what your recording needs: length, file type, speaker names or an English voice.

What happens to the recording

The audio is sent over HTTPS for transcription and translation. The audio, transcripts and any voice files are deleted after delivery, within 30 minutes at most, and nothing is used to train models. Your account keeps only the file name, length, languages and credits used. For recordings of other people, get their consent first, and again before copying anyone's voice into English. How cloud speech processing handles privacy covers the questions worth asking any service.

What it doesn't do

• No live or real-time translation: it works on finished files. For conversations as they happen, see real-time speech translation.
• No speaker names in the transcript. Add them yourself using the timestamps.
• No links: download the file first, then upload it.
• No certified translation. For court, immigration or official use you need a certified human translator; the machine translation is a working copy. Certified translation versus an AI transcript explains the difference.

Costs for three common recordings

Signing in with Google gives 100 free credits, enough for 100 minutes of English text or a 2-minute English voice track. Top-ups are charged in Indian rupees at the day's exchange rate through Cashfree Payments (UPI, cards or netbanking), about ₹96 for 1,000 credits. The exact cost is shown before you start, and failed jobs are refunded automatically.

Checking the English before you rely on it

1. Read names and places against the original transcript; machine translation often renders a name as an ordinary word.
2. Check every number, date and amount, especially where the original uses a different number style.
3. Mark idioms and jokes that came through literally.
4. For anything you will quote or publish, ask a speaker of the original language to review the passages that matter. Post-editing machine translation shows how to do that quickly.

Using the tool on this page

• The spoken language is detected automatically; you select the output language — the spoken language itself for an untranslated transcript, or another language for a translation. Files can be up to 2 hours long.
• Audio is sent over HTTPS in small chunks and deleted with its transcripts within 30 minutes of delivery (how files are protected).
• You can download a timestamped transcript, subtitles (SRT and VTT) and plain text. Every output, step and limit is explained on the audio translator page.