Doubao Music Generation Skill

Based on the Volcano Engine Doubao Music Generation API, a single script supports two modes:

• BGM Mode: Text description -> instrumental background music (no vocals)
• Song Mode: Prompt or lyrics -> complete vocal song

Prerequisites

1. Install Dependencies

pip install requests

2. Configure Authentication

# Edit .env and fill in your Volcano Engine AccessKey and SecretKey
export VOLC_ACCESS_KEY=$VOLC_ACCESS_KEY

# 火山引擎 Secret Access Key（必填）
export VOLC_SECRET_KEY=$VOLC_SECRET_KEY

Obtain your AccessKeyId and SecretAccessKey from the Volcano Engine Console -> Key Management.

1. Instrumental BGM Generation

# Basic BGM generation
python3 scripts/gen_music.py \
  --mode bgm \
  --text "Cafe light music, piano and guitar, relaxing and healing, 60 seconds" \
  --output cafe_bgm.mp3

# Specify duration and format
python3 scripts/gen_music.py \
  --mode bgm \
  --text "Epic cinematic soundtrack, string ensemble with brass, magnificent and awe-inspiring" \
  --duration 90 \
  --format wav \
  --output epic.wav

# Use structured segments (v5.0)
python3 scripts/gen_music.py \
  --mode bgm \
  --text "Electronic dance music, EDM style, energetic" \
  --segments '[{"Name":"intro","Duration":10},{"Name":"chorus","Duration":30},{"Name":"outro","Duration":10}]' \
  --output edm_bgm.mp3

# Enable prompt rewrite (improves results)
python3 scripts/gen_music.py \
  --mode bgm \
  --text "Sports video soundtrack" \
  --rewrite \
  --duration 60 \
  --output sport.mp3

2. Vocal Song Generation

# Generate song from prompt
python3 scripts/gen_music.py \
  --mode song \
  --prompt "A song about summer and the seaside, lighthearted and cheerful" \
  --genre "Pop" \
  --mood "Happy" \
  --gender Female \
  --output summer.mp3

# Generate song from lyrics (v4.3, fine-grained control)
python3 scripts/gen_music.py \
  --mode song \
  --lyrics "[verse]
I remember that day, the day we fell in love
Promised each other we'd never say goodbye
[chorus]
The day I gave my heart to you
You disappeared before my eyes" \
  --model-version v4.3 \
  --genre "Folk" \
  --mood "Sentimental/Melancholic/Lonely" \
  --gender Female \
  --timbre "Ethereal" \
  --output folk_song.mp3

# Read lyrics from file
python3 scripts/gen_music.py \
  --mode song \
  --lyrics-file my_lyrics.txt \
  --genre "Chinese Style" \
  --mood "Dreamy/Ethereal" \
  --output guofeng.mp3

# Fine-grained control (key/tempo/instruments)
python3 scripts/gen_music.py \
  --mode song \
  --prompt "Jazz-style urban life" \
  --model-version v4.3 \
  --genre "Jazz" \
  --mood "Chill" \
  --gender Male \
  --timbre "Magnetic" \
  --key "C" \
  --kmode "Major" \
  --tempo "Moderato" \
  --instrument "Saxophone,Acoustic_Piano,Bass" \
  --scene "Coffee Shop,Evening" \
  --output jazz_city.mp3

# English song
python3 scripts/gen_music.py \
  --mode song \
  --prompt "A song about chasing dreams and never giving up" \
  --model-version v4.3 \
  --genre "Pop Rock" \
  --mood "Inspirational/Hopeful" \
  --gender Male \
  --lang English \
  --output dream.mp3

Parameter Reference

Common Parameters

BGM-Specific Parameters

Song-Specific Parameters

For more enum values, see references/params.md.

Notes

• Short simple music (30s + minimal description) is likely to trigger copyright checks; it is recommended to enrich the description, add more parameters, or increase the duration
• Audio download links are time-limited; the script automatically downloads the file locally after generation
• BGM v5.0 does not require separate Genre/Mood parameters; describe them directly in the Text field
• Vocal songs v4.0 and v4.3 differ in the supported enum values for each field; see the parameter documentation for details