Initial commit: per-language Chatterbox TTS demo

README.md

@@ -1,5 +1,5 @@
 ---
-title: Chatterbox-Multilingual-TTS-AllLang-Demo
+title: Chatterbox-Multilingual-TTS-V3
 emoji: 🌐
 colorFrom: indigo
 colorTo: blue
@@ -8,17 +8,17 @@ sdk_version: 5.29.0
 python_version: "3.10"
 app_file: app.py
 pinned: false
-short_description: "Chatterbox TTS — language-agnostic AllLang model."
+short_description: "Chatterbox Multilingual TTS V3 (t3_mtl23ls_v3)."
 models:
-  - ResembleAI/Chatterbox-Multilingual-AllLang
-  - ResembleAI/Chatterbox-Multilingual-AllLang
+  - ResembleAI/chatterbox
+  - ResembleAI/chatterbox
 ---
 
-# Chatterbox Multilingual TTS — Language-Agnostic (AllLang)
+# Chatterbox Multilingual TTS V3
 
-Chatterbox TTS — language-agnostic AllLang model.
+Chatterbox Multilingual TTS V3 (t3_mtl23ls_v3).
 
-Powered by [`ResembleAI/Chatterbox-Multilingual-AllLang`](https://huggingface.co/ResembleAI/Chatterbox-Multilingual-AllLang).
-Base components reused from [`ResembleAI/Chatterbox-Multilingual-AllLang`](https://huggingface.co/ResembleAI/Chatterbox-Multilingual-AllLang).
+Powered by [`ResembleAI/chatterbox`](https://huggingface.co/ResembleAI/chatterbox).
+Base components reused from [`ResembleAI/chatterbox`](https://huggingface.co/ResembleAI/chatterbox).
 
 Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

app.py

@@ -81,9 +81,9 @@ def generate_tts_audio(
 with gr.Blocks() as demo:
     gr.Markdown(
         """
-        # Chatterbox Multilingual TTS — Language-Agnostic (AllLang)
-        Chatterbox TTS — language-agnostic AllLang model.
-        Powered by model [`ResembleAI/Chatterbox-Multilingual-AllLang`](https://huggingface.co/ResembleAI/Chatterbox-Multilingual-AllLang).
+        # Chatterbox Multilingual TTS V3
+        Chatterbox Multilingual TTS V3 (t3_mtl23ls_v3).
+        Powered by model [`ResembleAI/chatterbox`](https://huggingface.co/ResembleAI/chatterbox).
         """
     )
     with gr.Row():

chatterbox/src/chatterbox/tts.py

@@ -18,11 +18,11 @@ from .models.voice_encoder import VoiceEncoder
 from .models.t3.modules.cond_enc import T3Cond
 
 
-REPO_ID = "ResembleAI/Chatterbox-Multilingual-AllLang"
-BASE_REPO_ID = "ResembleAI/Chatterbox-Multilingual-AllLang"
-T3_FILENAME = "t3_alllang.safetensors"
-TOKENIZER_FILENAME = "mtl_tokenizer.json"
-T3_TEXT_VOCAB_SIZE = 2352
+REPO_ID = "ResembleAI/chatterbox"
+BASE_REPO_ID = "ResembleAI/chatterbox"
+T3_FILENAME = "t3_mtl23ls_v3.safetensors"
+TOKENIZER_FILENAME = "grapheme_mtl_merged_expanded_v1.json"
+T3_TEXT_VOCAB_SIZE = 2454
 
 
 def punc_norm(text: str) -> str: