Molnia or SaluteSpeech: which to pick for transcribing voice messages
SaluteSpeech is the largest transcription bot in Russian-language Telegram: by Telegram's own counter it had 527,668 users a month in August 2026. It is free and unlimited, and if that is enough for you — take it, there is nothing to decide here. The difference starts where recording length, output format and speed matter.
Line by line
| What we compare | SaluteSpeech | Molnia |
|---|---|---|
| What comes back | A flat transcript as long as the recording | The key point on the first line, topics as paragraphs; full text one tap away |
| Recording length | Short recordings, file up to 20 MB | Up to 3 hours, file up to 2 GB |
| What it accepts | Voice messages, audio, video notes | The same plus video files and recordings forwarded from other chats |
| Languages | Russian, English, Kazakh | Detected automatically; the summary comes back in the language of the recording |
| Price | Free, no limits | 20 minutes a month free, then $5 for 1,000 minutes |
| Bot audience | 527,668 people a month (August 2026) | Smaller: a younger project |
When SaluteSpeech is the honest choice
If what you need is the verbatim text, the recordings are short and you would rather not pay at all, SaluteSpeech closes the task and asks nothing in return. It is a showcase for Sber's speech technology rather than a business built on transcription, which is why it has no limits and probably never will. Arguing with a free unlimited service on price makes no sense, and we will not try.
Where Molnia does something else
We answer with the point rather than the transcript: the first line is what the recording was sent for, then the topics as paragraphs. An hour of audio reads in about half a minute instead of an hour. Then there is length: three hours and two gigabytes against short recordings, plus video files and forwarded audio that SaluteSpeech will not take. If your task is not «read all of it» but «understand what this is about», the difference shows on the first recording.
Recording length and formats
One recording can run up to three hours and two gigabytes. Besides voice messages Molnia takes video notes, audio files and video: mp3, m4a, wav, ogg, mp4, including anything forwarded from another chat or channel. The twenty-megabyte ceiling that bots on the ordinary Bot API trip over doesn't apply here — the file is fetched directly.
Frequently asked
Which is better, SaluteSpeech or Molnia?
It depends on the task. If you need a free verbatim transcript of a short recording — SaluteSpeech, it is free and unlimited. If you need the point of a long recording, a video or a forwarded file, and you need it fast — Molnia.
Is SaluteSpeech really completely free?
Yes, it has no limits. It is a showcase for Sber's speech recognition technology rather than a standalone business, so there is no reason for it to monetize transcription.
How much more popular is SaluteSpeech?
By Telegram's public counter it had 527,668 users a month as of August 2026 — roughly twelve times the next bot in the category.
How does the output text differ?
SaluteSpeech returns a flat word-for-word transcript. Molnia compresses the recording into meaning blocks: the key point on top, topics as paragraphs. The full verbatim text is available with a button if you need it.
Which files does Molnia accept?
Voice messages, video notes, audio and video: mp3, m4a, wav, ogg, mp4 — up to three hours and two gigabytes in a single recording, forwarded ones included.