ElevenLabs 技能

用于与 ElevenLabs API 交互以进行声音生成、音乐和语音管理的核心工具。

设置

请参阅SETUP.md以获取先决条件和设置说明。

模型

v3 音频标签（方括号，非XML/SSML）

[laughs], [chuckles], [sighs], [clears throat], [whispers], [shouts]
[excited], [sad], [angry], [warmly], [deadpan], [sarcastic]
[grumpy voice], [philosophical], [whiny voice], [resigned]
[laughs hard], [sighs deeply], [pause]

标签可置于文本任意位置。自由组合。v3深度理解情感语境。

输出格式

所有脚本均通过--format参数支持多种输出格式：

工具

1. 语音合成 (speech.py)

使用ElevenLabs语音进行文本转语音。

# Basic usage
python3 {baseDir}/scripts/speech.py "Hello world" -v <voice_id> -o output.mp3

# With format option
python3 {baseDir}/scripts/speech.py "Hello world" -v <voice_id> -o output.pcm --format pcm_44100

# With voice settings
python3 {baseDir}/scripts/speech.py "Hello" -v <voice_id> -o out.mp3 --stability 0.7 --similarity 0.8

2. 音效 (sfx.py)

生成音效和短音频片段。

# Generate a sound
python3 {baseDir}/scripts/sfx.py "Cinematic boom" -o boom.mp3

# Generate a loop
python3 {baseDir}/scripts/sfx.py "Lo-fi hip hop beat" --duration 10 --loop -o beat.mp3

# Different format
python3 {baseDir}/scripts/sfx.py "Whoosh" -o whoosh.pcm --format pcm_44100

3. 音乐生成 (music.py)

生成完整的音乐作品或器乐轨道。

# Generate instrumental intro
python3 {baseDir}/scripts/music.py --prompt "Upbeat 6s news intro sting, instrumental" --length-ms 6000 -o intro.mp3

# Generate background bed
python3 {baseDir}/scripts/music.py --prompt "Soft ambient synth pad" --length-ms 30000 -o bed.mp3

# High quality MP3
python3 {baseDir}/scripts/music.py --prompt "Jazz piano" --length-ms 10000 -o jazz.mp3 --output-format mp3_44100_192

4. 语音库 (voices.py)

列出可用语音及其ID。

# List voices
python3 {baseDir}/scripts/voices.py

# JSON output
python3 {baseDir}/scripts/voices.py --json

5. 语音克隆 (voiceclone.py)

从音频样本创建即时语音克隆。

安全：默认情况下，此脚本仅从以下位置读取文件：

• ~/.openclaw/elevenlabs/voiceclone-samples/

请将您的样本文件复制到该目录（或通过--sample-dir参数指定路径）。脚本会阻止读取样本目录之外的文件。

# Clone from audio files (put samples into ~/.openclaw/elevenlabs/voiceclone-samples)
python3 {baseDir}/scripts/voiceclone.py --name "MyVoice" --files sample1.mp3 sample2.mp3

# Use a custom sample dir
python3 {baseDir}/scripts/voiceclone.py --name "Andi" --sample-dir ./samples --files a.m4a b.m4a --language de --gender male

# With description and noise removal
python3 {baseDir}/scripts/voiceclone.py --name "Andi" --files a.m4a b.m4a --description "German male" --denoise

6. 配额与使用情况 (quota.py)

检查订阅配额及使用统计信息。

# Show current quota
python3 {baseDir}/scripts/quota.py

# Include usage breakdown by voice
python3 {baseDir}/scripts/quota.py --usage

# Last 7 days usage
python3 {baseDir}/scripts/quota.py --usage --days 7

# JSON output
python3 {baseDir}/scripts/quota.py --json

Output:

📊 ElevenLabs Quota
=======================================
Plan:      pro (active) — annual
Characters: 66.6K / 500.0K (13.3%)
           [███░░░░░░░░░░░░░░░░░░░░░░░░░░░]
Resets:    2026-02-18 (29 days)
Voices:    22 / 160 (IVC: ✓)
Pro Voice: 0 / 1 (PVC: ✓)