Add bark model

TTS/.models.json

@@ -9,6 +9,19 @@
                     "commit": "e9a1953e",
                     "license": "CC BY-NC-ND 4.0",
                     "contact": "egolge@coqui.ai"
+                },
+                "bark": {
+                    "description": "🐶 Bark TTS model released by suno-ai. You can find the original implementation in https://github.com/suno-ai/bark.",
+                    "hf_url": [
+                        "https://coqui.gateway.scarf.sh/bark/coarse_2.pt",
+                        "https://coqui.gateway.scarf.sh/bark/fine_2.pt",
+                        "https://coqui.gateway.scarf.sh/bark/text_2.pt",
+                        "https://coqui.gateway.scarf.sh/bark/config.json"
+                    ],
+                    "default_vocoder": null,
+                    "commit": "e9a1953e",
+                    "license": "MIT",
+                    "contact": "https://www.suno.ai/"
                 }
             }
         },
@@ -220,12 +233,12 @@
                     "license": "apache 2.0",
                     "contact": "adamfroghyar@gmail.com"
                 }
-
             },
             "multi-dataset": {
                 "tortoise-v2": {
                     "description": "Tortoise tts model https://github.com/neonbjb/tortoise-tts",
-                    "github_rls_url": ["https://coqui.gateway.scarf.sh/v0.14.1_models/autoregressive.pth",
+                    "github_rls_url": [
+                        "https://coqui.gateway.scarf.sh/v0.14.1_models/autoregressive.pth",
                         "https://coqui.gateway.scarf.sh/v0.14.1_models/clvp2.pth",
                         "https://coqui.gateway.scarf.sh/v0.14.1_models/cvvp.pth",
                         "https://coqui.gateway.scarf.sh/v0.14.1_models/diffusion_decoder.pth",

TTS/api.py

@@ -342,7 +342,7 @@ class TTS:
 
     def download_model_by_name(self, model_name: str):
         model_path, config_path, model_item = self.manager.download_model(model_name)
-        if isinstance(model_item["github_rls_url"], list):
+        if isinstance(model_item["model_url"], list):
             # return model directory if there are multiple files
             # we assume that the model knows how to load itself
             return None, None, None, None, model_path
@@ -580,6 +580,8 @@ class TTS:
                 Speed factor to use for 🐸Coqui Studio models, between 0.0 and 2.0. Defaults to None.
             file_path (str, optional):
                 Output file path. Defaults to "output.wav".
+            kwargs (dict, optional):
+                Additional arguments for the model.
         """
         self._check_arguments(speaker=speaker, language=language, speaker_wav=speaker_wav, **kwargs)
 

TTS/tts/configs/bark_config.py

@@ -5,11 +5,14 @@ from typing import Dict
 from TTS.tts.configs.shared_configs import BaseTTSConfig
 from TTS.tts.layers.bark.model import GPTConfig
 from TTS.tts.layers.bark.model_fine import FineGPTConfig
+from TTS.tts.models.bark import BarkAudioConfig
 from TTS.utils.generic_utils import get_user_data_dir
 
 
 @dataclass
 class BarkConfig(BaseTTSConfig):
+    model: str = "bark"
+    audio: BarkAudioConfig = BarkAudioConfig()
     num_chars: int = 0
     semantic_config: GPTConfig = GPTConfig()
     fine_config: FineGPTConfig = FineGPTConfig()
@@ -31,7 +34,7 @@ class BarkConfig(BaseTTSConfig):
     COARSE_SEMANTIC_PAD_TOKEN: int = 12_048
     COARSE_INFER_TOKEN: int = 12_050
 
-    REMOTE_BASE_URL = "https://dl.suno-models.io/bark/models/v0/"
+    REMOTE_BASE_URL = "https://huggingface.co/erogol/bark/tree/main/"
     REMOTE_MODEL_PATHS: Dict = None
     LOCAL_MODEL_PATHS: Dict = None
     SMALL_REMOTE_MODEL_PATHS: Dict = None

TTS/tts/layers/bark/inference_funcs.py

@@ -52,7 +52,7 @@ def load_voice(voice: str, extra_voice_dirs: List[str] = []):
         return semantic, coarse, fine
 
     if voice == "random":
-        return None, None
+        return None, None, None
 
     voices = get_voices(extra_voice_dirs)
     try:
@@ -183,7 +183,7 @@ def generate_text_semantic(
     assert isinstance(text, str)
     text = _normalize_whitespace(text)
     assert len(text.strip()) > 0
-    if history_prompt is not None or base is not None:
+    if all(v is not None for v in history_prompt) or base is not None:
         if history_prompt is not None:
             semantic_history = history_prompt[0]
         if base is not None:
@@ -327,7 +327,7 @@ def generate_coarse(
         model.config.COARSE_RATE_HZ / model.config.SEMANTIC_RATE_HZ * model.config.N_COARSE_CODEBOOKS
     )
     max_semantic_history = int(np.floor(max_coarse_history / semantic_to_coarse_ratio))
-    if history_prompt is not None or base is not None:
+    if all(v is not None for v in history_prompt) or base is not None:
         if history_prompt is not None:
             x_history = history_prompt
             x_semantic_history = x_history[0]
@@ -477,7 +477,7 @@ def generate_fine(
         and x_coarse_gen.min() >= 0
         and x_coarse_gen.max() <= model.config.CODEBOOK_SIZE - 1
     )
-    if history_prompt is not None or base is not None:
+    if all(v is not None for v in history_prompt) or base is not None:
         if history_prompt is not None:
             x_fine_history = history_prompt[2]
         if base is not None:
@@ -572,4 +572,4 @@ def codec_decode(fine_tokens, model):
     emb = model.encodec.quantizer.decode(arr)
     out = model.encodec.decoder(emb)
     audio_arr = out.detach().cpu().numpy().squeeze()
-    save_wav(path="test.wav", wav=audio_arr, sample_rate=model.config.sample_rate)
+    return audio_arr

TTS/tts/layers/bark/model.py

@@ -4,6 +4,7 @@ Much of this code is adapted from Andrej Karpathy's NanoGPT
 """
 import math
 from dataclasses import dataclass
+from coqpit import Coqpit
 
 import torch
 import torch.nn as nn
@@ -131,7 +132,7 @@ class Block(nn.Module):
 
 
 @dataclass
-class GPTConfig:
+class GPTConfig(Coqpit):
     block_size: int = 1024
     input_vocab_size: int = 10_048
     output_vocab_size: int = 10_048