aiaimodel.cards

Qwen-Audio-3.1-ASR-Flash

Audio modelCard summarized

An end-to-end offline speech recognition model designed for meeting transcription, content production, and call analysis

Source: Speech-to-text overview (Model Studio) and Qwen Cloud model page

Use it via API

$0.15 / 1M tok
Code exampleCompare

Where to call it

  • Alibaba Cloud Model Studioqwen-audio-3.1-asr-flash-filetrans

Supports

No feature list in the maker's docs.

Full feature support

Good for

  • Call-center recordings, podcasts and interviews
  • Meeting and classroom live transcription (streaming)
  • Chinese dialect-heavy audio

Watch out for

Overview

Alibaba · 01Qwen-Audio, front

Qwen ASR 3.1

An end-to-end offline speech recognition model designed for meeting transcription, content production, and call analysis

Typical price
$0.15 / 1M tok
File transcription (qwen-audio-3.1-asr-flash-filetrans), input
Max length
12 h per file (filetrans)
Languages
Chinese (Mandarin, Cantonese, Wu, Hokkien, Hakka, Gan, Xiang, Jin and many regional accents), English, Japanese, Korean, Vietnamese, Thai, Indonesian, Malay, Filipino, Hindi, Arabic, French, German, Spanish, Portuguese, Russian, Italian, Dutch, Swedish, Danish, Finnish, Norwegian, Greek, Polish, Czech, Hungarian, Romanian, Bulgarian, Croatian, Slovak and more
Modes
speech-to-text (file), speech-to-text (real-time streaming)
Released
—

Qwen-Audio-3.1-ASR-Flash is Alibaba's current recommended speech recognition family: an end-to-end model for multilingual and Chinese-dialect transcription with hot words, prompt context, speaker diarization, punctuation and text normalization, robust to noise. The filetrans variant handles files up to 12 hours; streaming and message variants serve live captions and voice input. Qwen Cloud model page; ASR overview

qwen-audio-3.1-asr-flash-filetrans
Typical price
$0.15 / 1M tok
Max length
12 h per file (filetrans)
Languages
Chinese (Mandarin, Cantonese, Wu, Hokkien, Hakka, Gan, Xiang, Jin and many regional accents), English, Japanese, Korean, Vietnamese, Thai, Indonesian, Malay, Filipino, Hindi, Arabic, French, German, Spanish, Portuguese, Russian, Italian, Dutch, Swedish, Danish, Finnish, Norwegian, Greek, Polish, Czech, Hungarian, Romanian, Bulgarian, Croatian, Slovak and more
Weights
Closed
Released
—
  • speech-to-text (file)
  • speech-to-text (real-time streaming)
  • speaker diarization
  • hot words
  • prompt context
  • punctuation and text normalization
  • dialect ASR/AST output control