SOTA Open Source TTS
Move flash whisper
Lengyuef4e8633363f1
diff --git a/speech_lm/datasets/whisper_vq.py b/speech_lm/datasets/whisper_vq.pyindex d95aebd..a9fffb4 100644--- a/speech_lm/datasets/whisper_vq.py+++ b/speech_lm/datasets/whisper_vq.py@@ -109,8 +109,8 @@ if __name__ == "__main__":from torch.utils.data import DataLoaderfrom transformers import GenerationConfig- from speech_lm.models.flash_whisper import FlashWhisperForConditionalGenerationfrom speech_lm.models.whisper_vq import WhisperVQ+ from speech_lm.modules.flash_whisper import FlashWhisperForConditionalGenerationdataset = WhisperVQDataset("filelists/whisper-vq.test.filelist")dataloader = DataLoader(diff --git a/speech_lm/models/whisper_vq.py b/speech_lm/models/whisper_vq.pyindex 84e239e..6d4aff6 100644--- a/speech_lm/models/whisper_vq.py+++ b/speech_lm/models/whisper_vq.py@@ -5,7 +5,7 @@ import torchfrom torch import nnfrom vector_quantize_pytorch import VectorQuantize-from speech_lm.models.flash_whisper import (+from speech_lm.modules.flash_whisper import (FlashWhisperEncoderLayer,FlashWhisperForConditionalGeneration,)diff --git a/speech_lm/models/flash_whisper.py b/speech_lm/modules/flash_whisper.pysimilarity index 100%rename from speech_lm/models/flash_whisper.pyrename to speech_lm/modules/flash_whisper.pydiff --git a/tools/whisper_asr.py b/tools/whisper_asr.pyindex 16ff305..d242e1e 100644--- a/tools/whisper_asr.py+++ b/tools/whisper_asr.py@@ -16,7 +16,7 @@ from loguru import loggerfrom transformers import WhisperProcessorfrom whisper.audio import load_audio, log_mel_spectrogram, pad_or_trim-from speech_lm.models.flash_whisper import FlashWhisperForConditionalGeneration+from speech_lm.modules.flash_whisper import FlashWhisperForConditionalGenerationRANK_STR = ""