القائمة
SSML cookbook for Azure Neural TTS (zh-CN). Voice selection, express-as styles, break / prosody templates, char→sec math, and word-boundary collection for word-level subtitle timing.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
Authors the Chinese spoken script for football data-observation videos in the match-insight-harness harness WRITE phase. v2 is AGENT-FIRST: a single LLM call freely writes the whole conclusion-first four-act script per match (podcast dual-host OR monologue single-host); a deterministic template is only the offline / LLM-failure fallback. Structure-agnostic — driven by a market- derived brief, never hard-coded report sections.
Verification strategies for ensuring rendered video stays in sync with the source AudioManifest. Tolerances, fingerprinting and frame-OCR techniques used by phases 06 / 08 / 10.
| name | azure-speech-ssml |
| description | SSML cookbook for Azure Neural TTS (zh-CN). Voice selection, express-as styles, break / prosody templates, char→sec math, and word-boundary collection for word-level subtitle timing. |
| Role | Voice | Style | Rate | Pitch | When to use |
|---|---|---|---|---|---|
| Anchor | zh-CN-XiaoxiaoNeural | chat deg 1.2 | +5% | +1st | 提问、引导、收束 |
| Analyst | zh-CN-YunxiNeural | narration-relaxed | +2% | +0st | 数据解读、策略描述、长句 |
Substitutions allowed: zh-CN-XiaoyiNeural (warmer Anchor), zh-CN-YunjianNeural
(sportscaster Analyst). Update tools/ssml.ts#VOICE and tts-cache invalidates
automatically.
Chinese mainland TTS averages 4.2 chars/sec at default rate, including punctuation pauses. Latin chars count as 0.5. With micro-breaks injected at "、"(120ms) "，"(160ms) "；"(220ms) "。！？"(280ms), real-world deviation is typically ±8%.
<speak version="1.0"
xmlns="http://www.w3.org/2001/10/synthesis"
xmlns:mstts="http://www.w3.org/2001/mstts"
xml:lang="zh-CN">
<voice name="zh-CN-XiaoxiaoNeural">
<mstts:express-as style="chat" styledegree="1.2">
<prosody rate="+5%" pitch="+1st">
这场比赛，<break time="160ms"/>赛前有哪些变量？
</prosody>
</mstts:express-as>
</voice>
</speak>
prosody.rate by 5% (+5% → +0%).+5% → +10%); cap at +20%.serious.<emphasis level="strong">关键词</emphasis>.const synth = new SpeechSynthesizer(cfg, AudioConfig.fromAudioFileOutput(out));
const boundaries: { text: string; offsetMs: number; durMs: number }[] = [];
synth.wordBoundary = (_s, e) => boundaries.push({
text: e.text,
offsetMs: Number(e.audioOffset) / 10_000,
durMs: Number(e.duration) / 10_000,
});
Boundaries drive subtitles.vtt word-level grouping (default 3 chars/cue).
PCM 24kHz mono 16-bit Riff WAV (SpeechSynthesisOutputFormat.Riff24Khz16BitMonoPcm).
Read duration deterministically from RIFF header without ffprobe.
sha256(voice + "\u241f" + ssml). Bumping voice list / prosody defaults
invalidates only the affected lines.