Can it transcribe a Russian video into English?

Not directly: the result is Russian, in Cyrillic. For English subtitles, transcribe with Output set to Subtitles, download the .srt and open it in the subtitle translator, which runs on-device in Chrome 138 or later and keeps the timing. Two machine steps compound each other’s errors, so correct the Russian first.

Why does my Russian .srt show garbled characters?

The file is UTF-8, which current players and editors read correctly. Some older players and Windows programs assume the Windows-1251 encoding for Cyrillic and show strings like РџСЂРёРІРµС‚ instead of Привет. Set the player’s subtitle encoding to UTF-8, or re-save the file from an editor that lets you pick one.

Does it work for Ukrainian or Belarusian speech?

No. Neither is in the language picker, and the Russian setting is not a substitute: it writes whatever it hears as Russian, so Ukrainian or Belarusian speech comes back as Russian-looking text with words respelled or replaced. Russian spoken with a Ukrainian or Belarusian accent is different: that is still Russian, and the Accurate model handles the accent better.

Does my file get uploaded?

No. The models and all processing run inside your browser on your own device. You can verify this yourself: open your browser’s DevTools, watch the Network tab, and process a file: there are no uploads. After the first visit most tools also work fully offline.

The models are OpenAI’s Whisper (tiny and base tiers). On clear speech they are very usable; heavy accents, crosstalk and noisy rooms will show errors. The Accurate tier is noticeably better on difficult audio.

Which languages work?

Fifteen, picked in the Language setting: English, Spanish, French, German, Italian, Portuguese, Dutch, Russian, Polish, Turkish, Japanese, Korean, Chinese, Arabic and Hindi. The decoder is conditioned on the one you pick, so it transcribes that language rather than approximating it in English — which also means picking the wrong one returns confident nonsense. The weights understand 99 languages; these are the ones we measured as good enough to offer.

How long does it take?

Transcription runs on your processor, so speed depends on the machine: expect around a quarter of the recording’s length, and less on the Fast model than on Accurate.

Russian Audio Transcription: Speech to Text

Turn Russian speech into text in your browser, in Cyrillic: drop a recording or a video and get a Russian transcript as .txt, or timed .srt and .vtt subtitles. Russian is one of the larger languages in Whisper’s training, so on clear speech the small on-device models get most words right. Two things to expect: ё usually comes back as е, as in most printed Russian, and unstressed vowels are spelled from context, so rare words and surnames deserve a second look. The recording stays on your machine, with no upload and no account.

How it works

Drop a Russian audio file or video. Russian is already selected as the spoken language.
Use Accurate (≈77 MB) for phone calls, several speakers or a noisy room; Fast (≈41 MB) for a quick draft. Set Output to Subtitles before running if you want timed captions.
Press Transcribe, correct names in place, and download .txt, .srt or .vtt, all saved as UTF-8 Cyrillic text.

Questions

Can it transcribe a Russian video into English?: Not directly: the result is Russian, in Cyrillic. For English subtitles, transcribe with Output set to Subtitles, download the .srt and open it in the subtitle translator, which runs on-device in Chrome 138 or later and keeps the timing. Two machine steps compound each other’s errors, so correct the Russian first.
Why does my Russian .srt show garbled characters?: The file is UTF-8, which current players and editors read correctly. Some older players and Windows programs assume the Windows-1251 encoding for Cyrillic and show strings like РџСЂРёРІРµС‚ instead of Привет. Set the player’s subtitle encoding to UTF-8, or re-save the file from an editor that lets you pick one.
Does it work for Ukrainian or Belarusian speech?: No. Neither is in the language picker, and the Russian setting is not a substitute: it writes whatever it hears as Russian, so Ukrainian or Belarusian speech comes back as Russian-looking text with words respelled or replaced. Russian spoken with a Ukrainian or Belarusian accent is different: that is still Russian, and the Accurate model handles the accent better.
Does my file get uploaded?: No. The models and all processing run inside your browser on your own device. You can verify this yourself: open your browser’s DevTools, watch the Network tab, and process a file: there are no uploads. After the first visit most tools also work fully offline.
How accurate is it?: The models are OpenAI’s Whisper (tiny and base tiers). On clear speech they are very usable; heavy accents, crosstalk and noisy rooms will show errors. The Accurate tier is noticeably better on difficult audio.
Which languages work?: Fifteen, picked in the Language setting: English, Spanish, French, German, Italian, Portuguese, Dutch, Russian, Polish, Turkish, Japanese, Korean, Chinese, Arabic and Hindi. The decoder is conditioned on the one you pick, so it transcribes that language rather than approximating it in English — which also means picking the wrong one returns confident nonsense. The weights understand 99 languages; these are the ones we measured as good enough to offer.
How long does it take?: Transcription runs on your processor, so speed depends on the machine: expect around a quarter of the recording’s length, and less on the Fast model than on Accurate.

Transcribe Russian Audio

How it works

Questions

Related tools