IndexTts2Checkpoints

Models & Types · AiDotNet.TextToSpeech.Latest

The files IndexTTS2’s release reads (infer_v2.py and ensure_models_available).

Constructors

Constructor Summary
IndexTts2Checkpoints()

Properties

Property Summary
bool TextToEmotion { get; set; } Whether to load the text-to-emotion model too (Qwen3, about 0.6B parameters).
string Directory { get; set; } The IndexTeam/IndexTTS-2 directory: gpt.pth, s2mel.pth, wav2vec2bert_stats.pt, feat1.pt, feat2.pt, bpe.model and qwen0.6bemo4-merge.
string SemanticCodec { get; set; } MaskGCT’s semantic codec, semantic_codec/model.safetensors (amphion/MaskGCT).
string SemanticModel { get; set; } w2v-BERT 2.0’s model.safetensors (facebook/w2v-bert-2.0).
string SpeakerEncoder { get; set; } CAM++’s campplus_cn_common.bin (funasr/campplus).
string Vocoder { get; set; } BigVGAN-v2’s bigvgan_generator.pt (nvidia/bigvgan_v2_22khz_80band_256x).