Quick TTS

Zero-config text-to-speech — give text, get an mp3 file. Handles natural-language voice selection ("用女声", "撒娇语气", "生气一点") and auto-inserts pacing breaks for long text. Use this whenever a user wants to quickly synthesize speech without configuring API details, or says things like "帮我把这段话合成语音", "用温柔御姐的声音读一下", "生成一段生气语气的音频".

scikkk@scikkk

Install

openclaw skills install @scikkk/quick-tts

SenseAudio TTS Quick

Zero-config speech synthesis. User gives text, you produce an mp3 file — no API knowledge required.

Voice Selection

Default voice: male_0004_a (儒雅道长, neutral).

Map natural-language requests to voice IDs before calling the API:

User says	voice_id
女声 / 女生 / 女的	`female_0006_a` (温柔御姐)
温柔御姐	`female_0006_a`
嗲嗲 / 台妹	`female_0033_a`
撒娇	`female_0033_c`
低落 / 难过 / 悲伤	`female_0033_d`
委屈	`female_0033_e`
生气 / 愤怒	`female_0033_f`
开心 / 高兴 / 活泼	`female_0033_b`
傲娇	`female_0027_f`
病娇	`female_0027_c`
妩媚 / 性感	`female_0027_e`
男声 / 男生	`male_0004_a`
沙哑 / 深情	`male_0018_a`
撒娇青年	`male_0023_a`
乐观 / 少年	`male_0026_a`
儿童 / 小孩 / 萌娃	`child_0001_b`

If the user names a voice not in this table, use male_0004_a and note the fallback.

Long Text Pacing

For text longer than 100 characters, insert <break time="300"/> after each 。 ！ ？ to improve natural rhythm:

text

原文：今天天气很好。我们出去走走吧。
处理后：今天天气很好。<break time="300"/>我们出去走走吧。<break time="300"/>

Do not insert breaks inside quoted speech or after mid-sentence punctuation like ，.

API Call

bash

curl -s -X POST https://api.senseaudio.cn/v1/t2a_v2 \
  -H "Authorization: Bearer $SENSEAUDIO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "SenseAudio-TTS-1.0",
    "text": "<TEXT>",
    "stream": false,
    "voice_setting": {
      "voice_id": "<VOICE_ID>"
    },
    "audio_setting": {
      "format": "mp3"
    }
  }' -o response.json

# Decode hex audio to mp3
jq -r '.data.audio' response.json | xxd -r -p > output.mp3

Check base_resp.status_code == 0 before decoding. On error, show base_resp.status_msg.

Output

After generating the file:

Save as a descriptive filename, e.g. speech_<timestamp>.mp3 or based on the first few words of the text.
Report:
- File path
- Audio duration: jq '.extra_info.audio_length' response.json (ms → seconds)
- Character count: jq '.extra_info.usage_characters' response.json
- Voice used

Example output: