Music generation

Generate music, songs, or sound effects from a prompt.

term-llm music "single bright xylophone ping" --duration 1

By default, music clips are:

• Saved to ~/Music/term-llm/ with timestamped filenames
• Generated with Venice elevenlabs-sound-effects-v2
• Returned as MP3

The command also reads the prompt from stdin when no positional prompt is supplied:

echo "cinematic whoosh" | term-llm music -o - > whoosh.mp3

Providers

term-llm music supports:

• venice: Venice async audio queue for music, songs, sound effects, and Venice-hosted TTS audio models
• elevenlabs: ElevenLabs /v1/music, /v1/music/stream, and /v1/music/detailed

Music flags

--provider, --model, --format, and --voice include shell completion candidates.

Examples

term-llm music "single bright xylophone ping" \
  --provider venice \
  --model elevenlabs-sound-effects-v2 \
  --duration 1

term-llm music "upbeat chiptune victory sting" \
  --provider venice \
  --model mmaudio-v2-text-to-audio \
  --duration 1 \
  --format wav

term-llm music "polished instrumental funk loop" \
  --provider elevenlabs \
  --duration 3 \
  --force-instrumental \
  --format mp3_44100_128

term-llm music "80s synth pop song" \
  --provider venice \
  --model minimax-music-v25 \
  --lyrics "Verse 1: Neon lights over the avenue" \
  --duration 60

term-llm music "quote a one second sound effect" \
  --provider venice \
  --model elevenlabs-sound-effects-v2 \
  --duration 1 \
  --quote

Venice music models

term-llm includes the Venice music/audio model catalog:

Venice options are model-specific. If a model does not support a supplied field, Venice returns the API error directly.

ElevenLabs music model

ElevenLabs currently documents one music model for the compose endpoints:

ElevenLabs prompt-mode duration is 3 to 600 seconds. So a literal one-second test is not accepted by the direct ElevenLabs music API; use --duration 3 there. Venice has one-second-capable sound-effect models.

ElevenLabs output formats:

alaw_8000, mp3_22050_32, mp3_24000_48, mp3_44100_32, mp3_44100_64, mp3_44100_96, mp3_44100_128, mp3_44100_192, opus_48000_32, opus_48000_64, opus_48000_96, opus_48000_128, opus_48000_192, pcm_8000, pcm_16000, pcm_22050, pcm_24000, pcm_32000, pcm_44100, pcm_48000, ulaw_8000, wav_8000, wav_16000, wav_22050, wav_24000, wav_32000, wav_44100, wav_48000.

JSON output

--json prints a single structured object to stdout after saving the file.

{
  "provider": "venice",
  "prompt": "single bright xylophone ping",
  "model": "elevenlabs-sound-effects-v2",
  "format": "mp3",
  "output": {
    "path": "/home/me/Music/term-llm/20260503-120000-single_bright_xylophone_ping.mp3",
    "mime_type": "audio/mpeg",
    "bytes": 17180
  }
}

Credentials and config

term-llm music reads Venice credentials from VENICE_API_KEY, music.venice.api_key, audio.venice.api_key, or the existing image.venice.api_key fallback.

ElevenLabs credentials are read from ELEVENLABS_API_KEY, XI_API_KEY, music.elevenlabs.api_key, audio.elevenlabs.api_key, or the configured providers.elevenlabs API key.

music:
  provider: venice
  output_dir: ~/Music/term-llm
  venice:
    api_key: $VENICE_API_KEY
    model: elevenlabs-sound-effects-v2
    format: mp3
  elevenlabs:
    api_key: $ELEVENLABS_API_KEY
    model: music_v1
    format: mp3_44100_128