aps-agent/video/gen_audio.py

"""从 narration.md 解析 SEG 分段，逐段调用 TTS 生成 MP3。"""
import os
import re
import subprocess
import sys
from pathlib import Path

WS = Path(__file__).resolve().parent
plugin_dir = (os.environ.get("APS_AUDIO_GENERATION_PLUGIN") or "").strip()
voice_id = (os.environ.get("APS_TTS_VOICE_ID") or "").strip()
if not plugin_dir or not voice_id:
    raise SystemExit("请配置 APS_AUDIO_GENERATION_PLUGIN 和 APS_TTS_VOICE_ID 后再生成配音")
PLUGIN = Path(plugin_dir).expanduser().resolve()
VOICE = voice_id

text = (WS / "narration.md").read_text(encoding="utf-8")
segments = re.findall(r"SEG (\d+)（[^）]*）\s*\n(.*?)(?=\n\nSEG |\Z)", text, re.S)
print(f"共解析 {len(segments)} 段")

for num, body in segments:
    out = WS / "audio" / f"seg{num}.mp3"
    if out.exists() and out.stat().st_size > 10_000:
        print(f"seg{num} 已存在，跳过")
        continue
    body = " ".join(body.split())
    result = subprocess.run(
        [sys.executable, str(PLUGIN / "scripts" / "audio_generation_tool.py"), "speech",
         "--text", body, "--voice-id", VOICE, "--output", str(out)],
        cwd=PLUGIN, capture_output=True, text=True, timeout=600,
    )
    ok = out.exists() and out.stat().st_size > 10_000
    print(f"seg{num}: {'OK' if ok else 'FAIL'} ({len(body)} 字)")
    if not ok:
        print(result.stdout[-500:], result.stderr[-500:])
        sys.exit(1)
print("全部配音生成完成")