Generates audio using Eleven Labs API: TTS narration, sound effects, voice cloning, music, transcription. Use for audio generation, voice synthesis, or sound effects.
Generate high-quality audio using the Eleven Labs API.
Requires ELEVEN_API_KEY environment variable and uv package manager.
All scripts are UV-based and self-contained. Run directly:
Convert text/markdown to speech with female voices.
scripts/narrate.py "Hello world" -o hello.mp3
scripts/narrate.py -f article.md -o article.mp3
scripts/narrate.py "Welcome!" -v rachel -o welcome.mp3
Voices: sarah (default), rachel, dorothy
Create one-shot audio effects from descriptions.
scripts/sound-effect.py "short notification ding" -o ding.mp3 -d 1.5
scripts/sound-effect.py "whoosh swoosh sound" -o whoosh.mp3
scripts/sound-effect.py "rain on window" -d 10 --loop -o rain.mp3
Clone a voice from audio samples (min 3 recommended).
scripts/clone-voice.py "My Voice" sample1.mp3 sample2.mp3 sample3.mp3
Transform audio to use a different voice.
scripts/voice-convert.py recording.mp3 -o sarah_version.mp3
scripts/voice-convert.py speech.wav -v rachel -o rachel_version.mp3
scripts/voice-convert.py noisy.mp3 -o clean.mp3 --remove-noise
Compose original music from prompts.
scripts/compose-music.py "upbeat electronic background music" -o bg.mp3
scripts/compose-music.py "calm piano melody" -d 60 -o piano.mp3
Remove background noise from recordings.
scripts/isolate-audio.py noisy_recording.mp3 -o clean.mp3
Transcribe audio with timestamps.
scripts/transcribe.py recording.mp3
scripts/transcribe.py podcast.mp3 -o transcript.json
scripts/transcribe.py video.mp4 --format srt -o subtitles.srt
Generate new synthetic voices from descriptions.
scripts/design-voice.py "A warm, friendly female voice with British accent"
scripts/design-voice.py "Deep male narrator" -t "Welcome to the show" -o preview.mp3
Female voices:
EXAVITQu4vr4xnSDxMaL — Sarah (soft, warm)21m00Tcm4TlvDq8ikWAM — Rachel (calm, clear)ThT5KcBeYPX3keUQqHPh — Dorothy (pleasant)eleven_flash_v2_5 — Fastest, 32 languages, lowest costeleven_multilingual_v2 — Best quality, 29 languageseleven_turbo_v2_5 — Balanced speed/qualitymp3_44100_128 — Good quality (default)mp3_22050_32 — Smallest file sizepcm_16000 — Raw PCM for processingulaw_8000 — Twilio compatible