ComfyUI-Index-TTS Dev
c76fe93f9a
- vendor upstream indextts infer_v2_5 code into indextts2_5/vendor - port transformers>=5 compatibility shims (_compat + patched beam search/generation utils) - local-first model paths under ComfyUI/models/IndexTTS-2.5 (same convention as IndexTTS-2) - 5 new nodes: Base / EmotionAudio / EmotionVector / EmotionText / CacheControl - TTS2_5_download.py + TTS2.5模型路径.txt following TTS2 conventions - engine-level tests: 12/12 passed; ComfyUI E2E workflow test passed
24 lines
907 B
Python
24 lines
907 B
Python
# -*- coding: utf-8 -*-
|
|
"""单独补测：显式情感描述（Qwen 转换）"""
|
|
import sys, os, time
|
|
sys.path.insert(0, os.path.abspath(os.path.dirname(__file__)))
|
|
import librosa
|
|
import soundfile as sf
|
|
from indextts2_5 import IndexTTS25Engine
|
|
|
|
OUT_DIR = os.path.join(os.path.dirname(os.path.abspath(__file__)), "test_outputs_2_5")
|
|
os.makedirs(OUT_DIR, exist_ok=True)
|
|
|
|
engine = IndexTTS25Engine()
|
|
ref, ref_sr = librosa.load("TimbreModel/林志玲.wav", sr=None, mono=True)
|
|
|
|
t0 = time.time()
|
|
sr, wav, sub = engine.generate(
|
|
text="快躲起来！是他要来了！他要来抓我们了！", lang="ZH",
|
|
reference_audio=(ref, ref_sr),
|
|
use_qwen=True, emo_text="你吓死我了！你是鬼吗？", emo_weight=0.6, verbose=True,
|
|
)
|
|
path = os.path.join(OUT_DIR, "emo_text_explicit.wav")
|
|
sf.write(path, wav, sr)
|
|
print(f"[PASS] emo_text_explicit: {len(wav)/sr:.2f}s -> {path}, cost={time.time()-t0:.1f}s")
|