Not the person who made your subtitles. “DimaTorzok” is a hallucination produced by the Whisper speech-to-text model, which invents a subtitle-author credit when the audio goes silent or noisy.

Why does Whisper write “Subtitles by DimaTorzok”?

Whisper was trained on huge amounts of video paired with subtitles, where author credits often appear over silence at the end. The model learned that silence equals a credit line, so it reproduces one on empty audio.

Did DimaTorzok hack or scam the AI?

No. That's a popular myth. There is no “digital fingerprint.” His subtitle credits simply ended up in the model's training data by accident.

How do I remove DimaTorzok from subtitles?

Delete the credit line manually, or run your text or subtitle file through a free cleaner that detects and removes known hallucination phrases while keeping timecodes.

Who Is DimaTorzok?

Short answer: not a person. “Subtitles by DimaTorzok” is a hallucination of an AI speech-to-text model (OpenAI's Whisper). When the audio falls silent or noisy, the model doesn't stay quiet — it “makes up” text, most often a subtitle-author credit that was never there. Nobody actually made those subtitles.

Remove DimaTorzok from my subtitles → Step-by-step guide

Where this phrase shows up

You may have seen it during automatic speech recognition in:

voice-message transcription in Telegram;
auto-captions in CapCut, Instagram Edits, DaVinci Resolve;
transcripts of interviews, podcasts and lectures made with Whisper;
generated .srt/.vtt subtitle files.

The common thread: at that moment there was a pause, silence or music in the recording.

Why the AI writes it

Whisper was trained on roughly 680,000 hours of audio paired with existing subtitles — including YouTube and pirated releases. Those videos often end with a line like “Subtitles by …” over silence. The model learned the pattern: silence at the end ≈ an author credit. So when it hits an empty stretch, it confidently writes what its experience says usually goes there.

This is called a model hallucination — invented text that wasn't in the audio. The same bug exists in other languages: Turkish “Altyazı M.K.”, Czech “Titulky vytvořil JohnyX”, Arabic “ترجمة نانسي قنقر”. Only the name in the credit changes.

Debunked: he did NOT scam the AI

A catchy version circulates online: that DimaTorzok “made AI treat silence as his digital fingerprint and scammed all the generators.” That's false.

There was no injection, no hack, no “fingerprint.” A real subtitler using the name DimaTorzok once made captions and signed his work — completely normal practice. His credits accidentally ended up in Whisper's training set among millions of others. Less dramatic than the myth, but that's exactly how it works: not malice, just uncleaned training data.

Was there a real DimaTorzok?

Apparently yes — a pseudonym of someone who voluntarily subtitled other people's videos and signed them. Because that signature reached the training data, the name now “lives” inside the model and surfaces for thousands of people worldwide. We describe a technical phenomenon and do not publish personal data or speculation about any individual (see disclaimer).

How to remove it

You can delete the credit line by hand, but if it keeps appearing it's easier to run your text or subtitle file through our free cleaner. It knows common hallucination phrases (in several languages) and removes them while preserving timecodes.

Open the subtitle cleaner →

FAQ

Is DimaTorzok a virus or a hack?

No. It's a harmless artifact of how speech-to-text models work. It doesn't damage files and isn't related to hacking — just extra text you need to delete.

Why did it appear specifically in my voice message?

Most likely there was a pause/silence at the start or end of the recording (e.g. you didn't stop recording right away). The model “drew in” a credit over that empty part.

Can I stop it from appearing?

Yes — on the recognition side, filtering out silence (VAD) before transcription and post-processing both help. If you make subtitles in an editor, it's easier to strip such lines automatically (see the cleaner) or use services that handle it for you.

Does this only happen in Russian?

No. Whisper hallucinates credits in many languages: “Subtitles by…”, “Altyazı M.K.” (TR), “Titulky vytvořil JohnyX” (CZ), “ترجمة نانسي قنقر” (AR) and more. Our cleaner catches a set of these signatures.