MiniMaxMiniMax ASR
MiniMax ASR 1.0
ASR 1.0 transcribes audio and can return speaker-labelled segments or subtitle files. It can also stream text output from an uploaded audio file.
Automatic language selection can be overridden
Without a language hint, ASR detects the dominant language and supports Chinese-English code switching. A language header can make the intended language explicit.MiniMax ASR 1.0 guide
Choose the output for the next step
Practical guidance. For a meeting record, speaker labels may be more useful than the first partial words. For captions, confirm segment timing against the recording. Break long recordings at sensible speech boundaries so that a sentence is not cut across requests.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | API |
| Model identifiers | asr-1.0 |
| Input limits | Up to 500 seconds and 50 MB per request. Oversized files are rejected rather than truncated.MiniMax ASR 1.0 guide |
| Output modes | Plain JSON, detailed JSON, SRT or VTT. Detailed timestamps and speaker labels cannot be combined with streaming output.MiniMax ASR 1.0 guide |
| Streaming boundary | The service streams transcription output; it does not accept continuous audio streaming input.MiniMax ASR 1.0 guide |
Related models
- MiniMax Speech 2.8
MiniMax · MiniMax Speech
A speech-generation model with native breath and laughter tags, 40 languages and HD or Turbo serving options.
Launch date API · Provider product