Implement VitsAudioConfig

TTS/tts/configs/vits_config.py

@@ -2,7 +2,7 @@ from dataclasses import dataclass, field
 from typing import List
 
 from TTS.tts.configs.shared_configs import BaseTTSConfig
-from TTS.tts.models.vits import VitsArgs
+from TTS.tts.models.vits import VitsArgs, VitsAudioConfig
 
 
 @dataclass
@@ -16,6 +16,9 @@ class VitsConfig(BaseTTSConfig):
         model_args (VitsArgs):
             Model architecture arguments. Defaults to `VitsArgs()`.
 
+        audio (VitsAudioConfig):
+            Audio processing configuration. Defaults to `VitsAudioConfig()`.
+
         grad_clip (List):
             Gradient clipping thresholds for each optimizer. Defaults to `[1000.0, 1000.0]`.
 
@@ -94,6 +97,7 @@ class VitsConfig(BaseTTSConfig):
     model: str = "vits"
     # model specific params
     model_args: VitsArgs = field(default_factory=VitsArgs)
+    audio: VitsAudioConfig = VitsAudioConfig()
 
     # optimizer
     grad_clip: List[float] = field(default_factory=lambda: [1000, 1000])

TTS/tts/models/vits.py

@@ -200,6 +200,18 @@ def wav_to_mel(y, n_fft, num_mels, sample_rate, hop_length, win_length, fmin, fm
     return spec
 
 
+#############################
+# CONFIGS
+#############################
+
+class VitsAudioConfig(Coqpit):
+    sample_rate: int = 22050
+    win_length: int = 1024
+    hop_length: int = 256
+    num_mels: int = 80
+    mel_fmin: int = 0
+    mel_fmax: int = None
+
 ##############################
 # DATASET
 ##############################