transcribe
Audio Transcription
Transcribe audio files from meetings into text documents.
Usage
/transcribe [filename]
/transcribe .research/meetings/audio/2024-12-02-lab-meeting.m4a
/transcribe .research/meetings/audio/ # Transcribe all untranscribed audio in directory
When to Use
- After recording a meeting, seminar, or discussion
- When RA detects new audio files in meetings/audio/ folder
- Before running /summarize_meeting
Supported Formats
- .m4a, .mp3, .wav, .webm, .mp4 (audio track)
- .ogg, .flac
Execution
The command runs:
conda run -n research-assistant python .ra/skills/transcribe/scripts/transcribe.py [filename or .research/meetings/audio/]
Behavior:
- If
[filename]provided: transcribe that specific audio file - If no filename (or
.research/meetings/audio/specified): automatically detect all audio files without transcripts and process them - If transcript already exists for a file: skip it
- Output saves to
.research/meetings/transcripts/[same-name].md
Post-Transcription Options
Transcription complete!
A) Run /summarize_meeting to extract action items and create tasks
B) Open transcript to review manually first
C) Continue with other work
What would you like to do?
Quality Notes
Improving Transcription Quality
- Use good microphone/recording quality
- Minimize background noise
- Speak clearly and at moderate pace
- Identify speakers at start if possible
Limitations
- Speaker diarization may be imperfect
- Technical terms may need manual correction
- Timestamps are approximate
Configuration
Environment variables (optional):
WHISPER_MODEL: Model size (default: "small", options: tiny, base, small, medium, large-v3)WHISPER_LANGUAGE: Force language (default: auto-detect)HF_TOKEN: HuggingFace token for speaker diarization
Related Skills
summarize-meeting- Extract action items from transcriptnext- Get next suggestion
Notes
- Raw transcripts may contain errors - review before citing
- Keep original audio files as source of truth
- Transcripts are for internal use, not publication
More from braselog/researchassistant
scientific-writing
Write scientific manuscripts with proper structure (IMRAD), citations (APA/AMA/Vancouver), figures/tables, and reporting guidelines (CONSORT/STROBE/PRISMA). Use when drafting any manuscript section, improving writing clarity, or preparing for journal submission.
15deep-research
Conduct a thorough literature search on a topic with verified citations. Use when the user types /deep_research, asks to "research a topic", "find papers on", or needs literature review. CRITICAL - Never fabricate citations. Every claim must have a verifiable source.
9statistical-analysis
Comprehensive statistical analysis toolkit for research. Conduct hypothesis tests (t-test, ANOVA, chi-square), regression, correlation, Bayesian stats, power analysis, assumption checks, and APA reporting. Use when the user asks about statistics, needs help analyzing data, or when writing methods sections that include statistical approaches.
5literature-review
Conduct comprehensive, systematic literature reviews using multiple databases (PubMed, bioRxiv, Semantic Scholar, OpenAlex). Creates documented searches, synthesizes findings thematically, verifies citations, and generates professional markdown reports with multiple citation styles (APA, Nature, Vancouver). Use when the user needs thorough literature research or types /deep_research.
5next
Assess current project state and suggest the most valuable next action. The primary entry point for users who aren't sure what to do. Analyzes context, checks for issues, and recommends specific skills or actions.
4quarterly-review
Conduct a quarterly review of your overall research mission and direction. This is a user-level review stored in ~/.researchAssistant/. Use when the user types /quarterly_review, every 3 months, after major project milestones, or when questioning research direction.
4