mm config¶

View and set extraction mode configuration — Whisper model, audio speed, beam size, and transcription backend settings stored in ~/.config/mm/mm.toml.

Synopsis¶

mm config SUBCOMMAND [OPTIONS]

Subcommands¶

Subcommand	Description
`show`	Display resolved configuration with current values
`init`	Create the default config file
`set KEY VALUE`	Set a single config key
`reset-db`	Delete all mm databases and caches
`reset-profiles`	Restore all profiles to built-in defaults
`reset`	Reset everything: databases and profiles

mm config show¶

Display the active configuration. Values are read from ~/.config/mm/mm.toml when it exists; built-in defaults are used otherwise.

mm config show
mm config show --format json
mm config show --format tsv

Options:

Flag	Description
`--format FORMAT`	Output format: `rich` (default), `json`, `tsv`, `csv`

Example output (rich):

Extraction Modes
  mode      whisper_model   audio_speed   beam_size
  ─────────────────────────────────────────────────
  fast      tiny            1.6           1
  accurate  medium          1.0           5

Transcription
  key        value
  ─────────────────────────────────
  backend    openai
  base_url   http://localhost:11434/v1
  api_key    ••••

mm config init¶

Create the default config file at ~/.config/mm/mm.toml. Does nothing if the file already exists unless --force is passed.

mm config init
mm config init --force    # overwrite existing config

Options:

Flag	Short	Description
`--force`	`-f`	Overwrite the existing config file

mm config set¶

Set a single configuration key. Changes are written to ~/.config/mm/mm.toml, creating the file if it does not exist.

mm config set KEY VALUE

Mode keys¶

Controls Whisper transcription behavior per extraction mode.

Key	Description	Example
`mode.fast.whisper_model`	Whisper model for fast mode	`tiny`, `base`, `small`
`mode.fast.audio_speed`	Audio playback speed multiplier for fast mode	`1.6`
`mode.fast.beam_size`	Whisper beam size for fast mode	`1`
`mode.accurate.whisper_model`	Whisper model for accurate mode	`medium`, `large-v3`
`mode.accurate.audio_speed`	Audio playback speed multiplier for accurate mode	`1.0`
`mode.accurate.beam_size`	Whisper beam size for accurate mode	`5`

Transcription keys¶

Controls the transcription backend used for audio and video.

Key	Description	Example
`transcription.backend`	Transcription backend	`openai`, `mlx`, `ctranslate2`
`transcription.base_url`	API base URL for `openai` backend	`http://localhost:11434/v1`
`transcription.api_key`	API key for the transcription endpoint	`sk-...`

Examples:

# Use a faster Whisper model for quick transcription
mm config set mode.fast.whisper_model tiny
mm config set mode.fast.audio_speed 2.0
mm config set mode.fast.beam_size 1

# Use a more accurate model for detailed transcription
mm config set mode.accurate.whisper_model large-v3
mm config set mode.accurate.audio_speed 1.0
mm config set mode.accurate.beam_size 5

# Point at a local OpenAI-compatible transcription server
mm config set transcription.backend openai
mm config set transcription.base_url http://localhost:11434/v1
mm config set transcription.api_key sk-...

mm config reset-db¶

Delete all mm databases and caches. This removes the SQLite database at ~/.local/share/mm/mm.db and any legacy cache files. This action is irreversible.

mm config reset-db
mm config reset-db --yes    # skip confirmation prompt

Options:

Flag	Short	Description
`--yes`	`-y`	Skip the confirmation prompt

This clears: - All cached LLM extractions (extractions table) - All chunked content and embeddings (chunks / chunks_vec tables) - File metadata index (files table)

mm config reset-profiles¶

Reset all profiles to built-in defaults. Removes custom profiles and restores reserved profiles to their default values. Mode settings (Whisper model, audio speed) are preserved.

mm config reset-profiles
mm config reset-profiles --yes    # skip confirmation

Options:

Flag	Short	Description
`--yes`	`-y`	Skip the confirmation prompt

mm config reset¶

Combines reset-db and reset-profiles into a single operation. Deletes all databases, clears all caches, and restores profiles to defaults. This action is irreversible.

mm config reset
mm config reset --yes    # skip all confirmation prompts

Options:

Flag	Short	Description
`--yes`	`-y`	Skip the confirmation prompt

Configuration file¶

mm reads configuration from ~/.config/mm/mm.toml. The file is optional; built-in defaults apply when it does not exist.

[mode.fast]
whisper_model = "tiny"
audio_speed = 1.6
beam_size = 1

[mode.accurate]
whisper_model = "medium"
audio_speed = 1.0
beam_size = 5

[transcription]
backend = "openai"
base_url = "http://localhost:11434/v1"
api_key = ""

Transcription backends¶

The transcription.backend key selects the Whisper runtime:

Backend	Description	Install
`openai`	Any OpenAI-compatible `/audio/transcriptions` endpoint (default)	Built-in
`mlx`	Apple Metal GPU via `lightning-whisper-mlx` — fastest on Apple Silicon	`pip install mm-ctx[mlx]`
`ctranslate2`	CPU int8 / CUDA float16 via `faster-whisper`	`pip install mm-ctx[gpu]`

The backend can also be overridden per-invocation with --encode.backend:

mm cat audio.mp3 -m accurate --encode.backend mlx
mm cat audio.mp3 -m accurate --encode.backend ctranslate2

Notes¶

mm config show sources its caption from the actual config file path, so you can verify which file is being read.
reset-db lists every file that will be deleted before prompting for confirmation.
reset-profiles preserves [mode.*] and [transcription.*] settings — only the [profiles] sections are affected.
The api_key field is always masked in show output and JSON (••••).