ASR Transcribe to Text (speaker-labeled)

Turns audio or video into speaker-labeled, timestamped transcripts — locally with MLX/whisper.cpp on Apple Silicon or through a remote ASR endpoint.

Image & VideoAdvanced★ 1,425⑂ 219AI score 9/10Last updated: Sep 23, 2026

What it does

Takes local audio/video files, direct media URLs, or podcast/web pages and produces [start-end] SPEAKER_00: text TXT plus a CSV.
Long recordings (>30 min) run a checkpointed whisper.cpp + Silero VAD route on the original timeline, then late-fuse pyannote speaker segments; interrupted runs resume from cached blocks.
Short/medium recordings use Qwen3-ASR (MLX) with mlx-whisper word timestamps; remote vLLM text can be aligned locally via --text-file.
Pre-ASR prep is built in: merge recorder segments in order, normalize to 16 kHz mono, pitch-preserving speedup to cut metered minutes, format selection (wav/m4a/ogg/flac) and self-verification of the output.
Includes guardrails against re-transcribing audio that already has a canonical transcript, plus an optional Feishu Minutes upload branch.

Who it's for

Anyone producing meeting, interview, or field-recording transcripts that must be attributed per speaker.
macOS Apple Silicon users who need offline/private transcription.
Engineers running self-hosted vLLM/ASR endpoints who need reliable long-form and batch handling.

Example uses

Feed a 2-hour meeting WAV and get a checkpointed, speaker-labeled TXT/CSV with timestamps.
Merge several 30-minute DJI mic segments into one file and export a 1.3x-speed M4A to cut billed minutes on a per-minute cloud ASR.
Pass a podcast episode URL to download the media and get plain text only (--no-diarization) for summarizing.

· · · Install guide · · ·

Try it now, no install

Paste this into Claude to use the skill without installing anything.

Read the instructions in this file and follow them to help me:
https://raw.githubusercontent.com/daymade/claude-code-skills/HEAD/daymade-audio/asr-transcribe-to-text/SKILL.md

What I want: (describe your task here)

If Claude can't open the link, open it yourself and paste the contents instead.

↓ If it works for you, download the ZIP below and install it. Then it runs on its own — no pasting each time.

Install in the Claude app (no terminal)

Download the ZIP with the button below.
In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.

↓ Download ZIP

Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the daymade-audio/asr-transcribe-to-text folder from the GitHub repo daymade/claude-code-skills into my ~/.claude/skills/asr-transcribe-to-text/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/daymade/claude-code-skills.git && mkdir -p ~/.claude/skills && cp -r claude-code-skills/daymade-audio/asr-transcribe-to-text ~/.claude/skills/

⚠ This is a third-party skill. Check the source repository before installing.

Open a terminal and cd into a working directory.
Clone the repo: git clone https://github.com/daymade/claude-code-skills.git
Copy the skill in place: mkdir -p ~/.claude/skills && cp -r claude-code-skills/daymade-audio/asr-transcribe-to-text ~/.claude/skills/
Install the required tooling: brew install ffmpeg uv (ffmpeg, ffprobe and uv are all used).
For speaker labels, accept the terms at https://hf.co/pyannote/speaker-diarization-3.1 and run huggingface-cli login (or export HF_TOKEN) once.
If you plan to use the long-recording path, download the whisper.cpp binary, a ggml large-v2/v3 model and the Silero VAD model, and note their absolute paths.
Restart Claude Code and ask something like "transcribe this recording with speaker labels", passing the file path. On first run answer the local-vs-remote setup questions; the config is saved for later runs.

View source on GitHub ↗License: MIT