ReleaseQwen (Alibaba Cloud)Qwen (Alibaba Cloud)published Sep 22, 2025seen 5d

QwenLM/Qwen3-ASR-Toolkit v1.0.4

QwenLM/Qwen3-ASR-Toolkit

Open original ↗

Captured source

source ↗
published Sep 22, 2025seen 5dcaptured 15hhttp 200method plain

V1.0.4

Repository: QwenLM/Qwen3-ASR-Toolkit

Tag: v1.0.4

Published: 2025-09-22T10:38:51Z

Prerelease: no

Release notes:

Release v1.0.4

This release introduces major enhancements for video content creators and power users by adding support for generating timestamped SRT subtitle files and providing greater control over the audio splitting process.

🚀 New Features

  • SRT Subtitle Generation: You can now generate industry-standard .srt subtitle files, complete with timestamps for each transcribed segment. This is perfect for adding captions to videos or for any workflow that requires synchronized text. Simply use the --save-srt (or -srt) flag to enable this feature.
  • Customizable Segment Duration: Gain more control over the VAD-based audio splitting. With the new --vad-segment-threshold (or -d) argument, you can define the target duration (in seconds) for each audio chunk. This allows for creating more granular subtitles (by using a shorter duration) or optimizing for different types of audio content. The default is 120 seconds.

Notability

notability 3.0/10

Minor toolkit release, no traction