Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →zxkane/audio-transcribe
This skill should be used when the user explicitly asks to "transcribe a meeting", "transcribe audio", "transcribe a meeting recording", "convert audio to text", "generate meeting minutes from audio", "do speech-to-text", "transcribe with speaker diarization", "identify speakers in audio", "transcribe Chinese audio", "transcribe English audio", "transcribe Japanese audio", "multi-speaker transcription", "transcribe a podcast", "transcribe podcast episode", "transcribe an interview", "convert pod
Passed 17 of 21 test cases with the skill, and 5 of 21 without it.
and 15 more in the full report
14 fixed4 passed either way3 still failing1 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 14 test cases the skill fixed. Same model, same prompt, with and without the skill.
I want to enable Bedrock LLM cleanup for `meeting.flac` using Anthropic Claude Sonnet 4.6 in an AWS global profile deployment. What `--provider` and `--model` CLI values should I supply?
| Checked | Without skill | With skill |
|---|---|---|
| Sets --provider bedrock | Passed | Passed |
| Uses global.anthropic.claude-sonnet-4-6 or us.anthropic.claude-sonnet-4-6 for --model | Missed | Passed |
The instructions your model reads when the skill loads.
Transcribe multi-speaker audio into structured Markdown with automatic speaker diarization, hotword biasing, and optional LLM cleanup. Two ASR engine families are available: FunASR (Paraformer / SenseVoice / Whisper — fast, cheap, GPU or CPU, 99 languages) and MiMo-V2.5-ASR (Xiaomi's 8B model, local GPU only, stronger on proper nouns and code-switching). Both share the same VAD + speaker-clustering stack.
All scripts run directly from the plugin directory — no copying needed. Define this shorthand at the start of every session:
SCRIPTS=${CLAUDE_PLUGIN_ROOT}/skills/audio-transcribe/scripts
--lang | Model | Languages | Hotword |
|---|---|---|---|
zh (default) | SeACo-Paraformer | Chinese (CER 1.95%) | Yes |
zh-basic | Paraformer-large | Chinese | No |
en | Paraformer-en | English | No |
auto | SenseVoiceSmall | Auto-detect: zh/en/ja/ko/yue | No |
whisper | Whisper-large-v3-turbo | 99 languages | No |
mimo | MiMo-V2.5-ASR (local 8B, GPU-only) | zh/en/code-switch/dialects | No |
All presets include speaker diarization (CAM++) and VAD (FSMN).
mimo reuses the FSMN VAD + CAM++ stack around MiMo's text output.
Diarization caveat:
autoandwhisperdo not output per-sentence timestamps, so speaker diarization does not work with these presets. Usezh,zh-basic,en, ormimowhen speaker identification is needed (e.g., podcasts, meetings).
Before starting transcription, always ask the user:
test cases passed, vs 5 / 21 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.