Qwen ASR (C-based Offline)

PassAudited by ClawScan on May 1, 2026.

Overview

This is a purpose-aligned offline speech-to-text wrapper, but it relies on undeclared local tools and an unpinned external qwen-asr checkout/model that users should verify before use.

Before installing, verify the qwen-asr repository and model source, install trusted local dependencies such as FFmpeg/OpenBLAS, and be aware that invoking the skill will execute local conversion and ASR binaries on the audio file you provide.

Findings (3)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

What this means

If the external checkout or downloaded model is not the intended trusted version, the skill could build or run different local code than the user expects.

Why it was flagged

The core ASR implementation is expected to come from an external GitHub checkout and may be built locally, but the artifacts do not pin a commit or provide hashes for the repo/model.

Skill content
Run: git clone https://github.com/antirez/qwen-asr ~/.openclaw/workspace/qwen-asr ... make blas
Recommendation

Clone only from a trusted source, pin or verify the exact commit and model files, and consider building the binary manually before invoking the skill.

What this means

The skill will run local ffmpeg and qwen_asr binaries against the selected audio file.

Why it was flagged

The runner invokes local command-line tools to convert audio and run inference, which is expected for an ASR wrapper but still means invoking the skill executes local binaries.

Skill content
ffmpeg -y -i "$AUDIO" -ar 16000 -ac 1 -f wav ... "$BIN" "${ARGS[@]}"
Recommendation

Use trusted local binaries, keep the audio path user-selected, and declare ffmpeg/make/qwen_asr requirements so users know what will execute.

What this means

Users may be surprised that the skill needs FFmpeg despite the no-FFmpeg-dependency claim.

Why it was flagged

The documentation simultaneously says there is no FFmpeg dependency and describes FFmpeg-based preprocessing, which can mislead users about required local tooling.

Skill content
无 Python/GIL/FFmpeg 依赖 ... 仅支持 `.ogg`/`.mp3`/`.wav`→`.wav` 预处理(FFmpeg 内置支持)
Recommendation

Clarify that the ASR engine is C-based but the wrapper uses FFmpeg for audio conversion, and list FFmpeg in requirements if it is needed.