cli-transcribe

cli-transcribe is a Node.js CLI for transcription workflows across hosted and local providers.

Supported providers:

• assemblyai
• elevenlabs
• whispercpp

Implemented workflow commands:

• transcribe meeting
• transcribe notes
• transcribe youtube

The CLI also includes configuration, setup, status, provider, and model inspection commands.

Install

Install globally from npm:

npm install -g @fredlackey/cli-transcribe

Then run the CLI:

transcribe --help

For local development from source:

npm install
node src/bin/transcribe.js --help

Command Surface

Top-level commands:

transcribe configure
transcribe provider
transcribe media
transcribe transcript
transcribe job
transcribe model
transcribe meeting
transcribe notes
transcribe setup
transcribe status
transcribe youtube

Global options:

• --json
• --interactive
• --output <mode>
• --profile <name>
• --provider <provider>
• --model <model>
• --region <region>
• --debug

Current implementation notes:

• meeting, notes, youtube, configure, setup, status, provider, and model are implemented.
• media, transcript, and job are scaffolded command groups and currently return placeholder responses rather than full workflows.

Configuration

Config is stored at:

~/.config/cli-transcribe/config.json

Initialize config:

transcribe configure init

Inspect config:

transcribe configure list
transcribe configure get
transcribe configure get workflows.meeting
transcribe configure get workflows.note

Set values directly:

transcribe configure set workflows.meeting.defaultProvider assemblyai
transcribe configure set workflows.meeting.sourceDir /Users/flackey/Recordings
transcribe configure set workflows.meeting.outputDir /Users/flackey/Transcriptions
transcribe configure set workflows.note.defaultProvider elevenlabs
transcribe configure set workflows.note.sourceDir /Users/flackey/VoiceNotes
transcribe configure set workflows.note.outputDir /Users/flackey/NotesTranscripts
transcribe configure set workflows.youtube.downloadDir /Users/flackey/Downloads/cli-transcribe
transcribe configure set providers.local.whispercpp.modelPath /absolute/path/to/ggml-small.en.bin

Interactive configuration:

transcribe configure
transcribe configure providers
transcribe configure workflows

Hosted providers use saved provider accounts in config or a direct --key passed at runtime. whispercpp does not use an API key.

Setup And Status

Inspect or install local dependencies:

transcribe setup --check
transcribe setup
transcribe setup --dry-run

Current dependency targets:

• ffmpeg
• yt-dlp
• whisper-cli

Inspect runtime state:

transcribe status
transcribe --json status
transcribe --interactive status

transcribe status reports current config, local dependency readiness, provider setup, whispercpp model configuration, and workflow defaults.

Usage

Meeting transcription:

transcribe meeting --provider assemblyai --file /path/to/meeting.mp4
transcribe meeting --provider elevenlabs --file /path/to/meeting.wav --format markdown
transcribe meeting --provider whispercpp --file /path/to/meeting.m4a --out ./meeting.json --format json
transcribe meeting --provider assemblyai --file /path/to/stereo-call.wav --multichannel
transcribe meeting --provider assemblyai --folder /Users/flackey/Recordings
transcribe meeting --provider whispercpp --folder /Users/flackey/Recordings --out ./meeting-transcripts

Notes transcription:

transcribe notes --provider assemblyai --file /path/to/note.m4a
transcribe notes --provider elevenlabs --file /path/to/voice-memo.wav --format markdown
transcribe notes --provider whispercpp --file /path/to/brainstorming-note.m4a --out ./note.json --format json
transcribe notes --provider elevenlabs --folder /Users/flackey/VoiceNotes
transcribe notes --provider whispercpp --folder ./voice-notes --out ./note-transcripts

YouTube transcription:

transcribe youtube "https://youtube.com/watch?v=123" --provider assemblyai
transcribe youtube "https://youtube.com/watch?v=123" --provider whispercpp --speakers
transcribe youtube "https://youtube.com/watch?v=123" --provider elevenlabs --format markdown
transcribe youtube "https://youtube.com/watch?v=123" --provider assemblyai --keep-media

Usage notes:

• meeting and notes require exactly one of --file or --folder.
• In folder mode, existing transcript outputs are skipped automatically.
• When --out is provided in folder mode, it must be a directory.
• meeting auto-prepares .mp4 inputs by creating a sibling .mp3 with ffmpeg when needed.
• youtube --speakers and youtube --meeting are equivalent.

If you have a saved provider account slug, you can use it instead of the raw provider id:

transcribe meeting --provider my-assembly --file /path/to/meeting.mp4

If you do not want to use saved config, pass a key directly:

transcribe meeting --provider assemblyai --file /path/to/meeting.mp4 --key YOUR_API_KEY

Provider Notes

assemblyai:

• hosted async transcription flow
• supports meeting, notes, and youtube
• uses configured account credentials or --key

elevenlabs:

• hosted batch transcription flow
• supports meeting, notes, and youtube
• uses configured account credentials or --key

whispercpp:

• local offline transcription flow
• supports meeting, notes, and youtube
• requires a resolvable whisper-cli binary, ffmpeg, and a local model file
• model selection is controlled by config or global --model

Documentation

• COMMANDS.md
• docs/COMMANDS.md

Contributing

If you find a gap or have a feature request, open an issue or submit a pull request on GitHub.

Questions?

If you have questions, comments, or just want to talk shop, feel free to reach out.

Fred Lackey
[email protected]
https://fredlackey.com

License

Apache-2.0