Skip to content

CLI

Jump to bottom
Garon edited this page Jun 1, 2026 · 2 revisions

CLI

Glimpse ships a small command-line tool for running transcription from the terminal. It's handy for batching files, scripting, or just trying a model without opening the app.

Quick Issues

Install the CLI

The glimpse command isn't there by default. You install it from inside the app.

  1. Open Glimpse.
  2. Go to Settings → About.
  3. Under Advanced, find Command line and click Install CLI.

That puts glimpse on your PATH. The CLI is part of Glimpse Personal / commercial, so you'll need an active license (or the trial) for the install button to be enabled.

To remove it later, the same row shows Uninstall.

How do I run it?

glimpse

You should see a usage line listing three subcommands: models, transcribe, and serve.

If glimpse isn't found, the install step above didn't finish, or your shell needs a restart so it picks up the new PATH.

Where do flags go?

Two global flags can sit anywhere:

  • --cache-dir <DIR>, where models live
  • --json, print structured output

Everything else belongs to a subcommand, so it has to come after the subcommand name.

glimpse transcribe audio.wav --model whisper_large_v3_turbo_q8

Not:

glimpse --model … transcribe audio.wav

Manage models

List what's available:

glimpse models list

Install one:

glimpse models install whisper_large_v3_turbo_q8

Check on it:

glimpse models status whisper_large_v3_turbo_q8

Remove it:

glimpse models delete whisper_large_v3_turbo_q8

Transcribe a file

glimpse transcribe audio.wav --model whisper_large_v3_turbo_q8

Useful flags:

  • --language en, hint the language (auto-detected if omitted)
  • --prompt "...", initial prompt / context
  • --dictionary term, bias toward a term (repeat for more)
  • --timestamps, include segment timestamps
  • --response-format text|json|verbose_json|srt|vtt, defaults to text

Example with subtitles:

glimpse transcribe meeting.wav \
  --model whisper_large_v3_turbo_q8 \
  --timestamps \
  --response-format srt

Bring your own Whisper model

--model accepts either a curated model id (from models list) or a path to any Whisper ggml/GGUF file.

By absolute path:

glimpse transcribe audio.wav \
  --model /Users/you/Downloads/ggml-large-v3-turbo-q5_0.bin

Or by file name inside --cache-dir:

glimpse --cache-dir /Users/you/whisper-models \
  transcribe audio.wav \
  --model ggml-large-v3-turbo-q5_0.bin

Parakeet and Nemotron still have to be installed via models install. Only Whisper supports loose files for now.

Run it as a local API

glimpse serve --model whisper_large_v3_turbo_q8

Defaults to http://127.0.0.1:11435. Endpoints:

  • GET /v1/models
  • POST /v1/models/{id}/install
  • POST /v1/audio/transcriptions (OpenAI-compatible)

Common options:

  • --host / --port, change the bind address
  • --api-key <KEY>, require an API key on requests
  • --cors, allow browser clients
  • --remote-endpoint <URL>, proxy transcriptions to an OpenAI-compatible upstream instead of running locally
  • --remote-api-key / --remote-model, credentials and model for the upstream

Cache directory

Default location:

  • Mac: ~/Library/Application Support/com.glimpse.data/models
  • Windows: %APPDATA%\com.glimpse.data\models

Override per-command with --cache-dir, or globally with the GLIMPSE_SPEECH_CACHE_DIR env var.

Audio formats

transcribe reads 16 kHz mono PCM WAV directly. Anything else (mp3, m4a, mp4, mov, etc.) gets converted automatically via FFmpeg, so you need ffmpeg on PATH for non-WAV files.

Quick check:

ffmpeg -version

If that fails, see FFmpeg Mac or ffmpeg windows.

You can also pre-convert yourself:

ffmpeg -i input.mov -ar 16000 -ac 1 -c:a pcm_s16le output.wav

JSON output

Add --json (global) to get structured output from models list, models status, and models install. For transcribe, use --response-format json or verbose_json instead. verbose_json includes segments.

Still not working?

  • Make sure you ran Install CLI from Settings → About first.
  • Make sure flags are after the subcommand, not before.
  • For non-WAV audio, confirm ffmpeg -version works.
  • If a model "isn't installed," run glimpse models status <id> to see what's missing.

Still stuck? Open an issue with the exact command and error.

Getting Started

Features

Help

Clone this wiki locally