Scribe is the speech-to-text model from ElevenLabs, the company best known for synthetic voices. You give it audio or video and get text with word timestamps, speaker diarization and audio-event tags such as laughter or music. The key point for Persian speakers is that Persian is in its list of supported languages.
What is Scribe good for?#
- Transcribing interviews and podcasts: turn long audio into text with speakers separated.
- Video subtitles: precise word timestamps are the basis for building subtitles.
- Meetings and calls: convert a recording to text and separate it by speaker.
- Software products: a developer adds speech-to-text to their own app via the API.
- Audio content analysis: use the text for summarization or searching an audio archive.
Strengths and weaknesses#
Strengths:
- Official support for Persian among dozens of languages.
- Word timestamps, speaker diarization and audio-event tags.
- An API suited to integrating into products.
- A real-time version for live speech.
Weaknesses:
- Accuracy figures are the company's own and must be tested on your audio.
- The interface is closer to a developer or creator tool than a ready meeting bot.
- Sign-up and payment from Iran are likely a problem.
- Dialects, noise and overlapping speakers reduce accuracy.
Free version and pricing#
ElevenLabs has a free tier with limits, and heavier use needs a subscription or usage-based payment. Check the official site for current prices and limits.
Language support and access from Iran#
Persian is in Scribe's language list, and you should check quality on your own audio, especially with regional dialects. Access from Iran is difficult: the company lists Iran among sanctioned countries in its terms of use, and sign-up and payment likely need a VPN and an international payment method.
