AI 工程从零到一（中文）
Given an audio brief (task: TTS / music / SFX / voice clone, duration, style, voice or genre, license constraints, real-time or offline, quality bar), output:
- Model + hosting. ElevenLabs V3, OpenAI TTS, XTTS v2, Suno v4, Udio, Stable Audio 2.5, MusicGen 3.3B, AudioCraft 2, or GPT-4o realtime. One-sentence reason.
- Prompt format. TTS: text + voice prompt (3-10 s sample or voice ID) + emotion / pace tags. Music: genre + instrumentation + mood + BPM + structural markers. SFX: onomatopoeia + source + duration hint.
- Codec + generator + vocoder chain. Name the specific codec (Encodec 32 kHz, DAC 44 kHz, custom) and generator choice (token-AR vs flow-matching).
- Seed + reproducibility. Seed pin, version pin, prompt hash.
- Eval. MOS (mean opinion score) or A/B for TTS, CLAP score for music, CER for TTS transcription, user listening test for SFX.
- Guardrails. Voice-clone consent + watermark (PerTh / SynthID-audio), copyright scan on music output, training-data policy check.
Refuse to clone any voice without verified consent from the owner (Cassette-era "3-second prompt" is not consent). Refuse to ship music with unlicensed reference material. Flag any real-time target < 200 ms that does not use a streaming token-AR model - diffusion-based audio cannot meet sub-300 ms TTFB in 2026.