commit files to HF hub

added_tokens.json

@@ -0,0 +1,4 @@
+{
+  "<ctc_blank>": 80,
+  "<mask>": 79
+}

config.json

@@ -0,0 +1,101 @@
+{
+  "_name_or_path": "microsoft/speecht5_tts",
+  "activation_dropout": 0.1,
+  "apply_spec_augment": true,
+  "architectures": [
+    "SpeechT5ForTextToSpeech"
+  ],
+  "attention_dropout": 0.1,
+  "bos_token_id": 0,
+  "conv_bias": false,
+  "conv_dim": [
+    512,
+    512,
+    512,
+    512,
+    512,
+    512,
+    512
+  ],
+  "conv_kernel": [
+    10,
+    3,
+    3,
+    3,
+    3,
+    2,
+    2
+  ],
+  "conv_stride": [
+    5,
+    2,
+    2,
+    2,
+    2,
+    2,
+    2
+  ],
+  "custom_pipelines": {
+    "text-to-speech": {
+      "impl": "tts.TTSPipeline",
+      "pt": [
+        "SpeechT5ForTextToSpeech"
+      ],
+      "tf": []
+    }
+  },
+  "decoder_attention_heads": 12,
+  "decoder_ffn_dim": 3072,
+  "decoder_layerdrop": 0.1,
+  "decoder_layers": 6,
+  "decoder_start_token_id": 2,
+  "encoder_attention_heads": 12,
+  "encoder_ffn_dim": 3072,
+  "encoder_layerdrop": 0.1,
+  "encoder_layers": 12,
+  "encoder_max_relative_position": 160,
+  "eos_token_id": 2,
+  "feat_extract_activation": "gelu",
+  "feat_extract_norm": "group",
+  "feat_proj_dropout": 0.0,
+  "guided_attention_loss_num_heads": 2,
+  "guided_attention_loss_scale": 10.0,
+  "guided_attention_loss_sigma": 0.4,
+  "hidden_act": "gelu",
+  "hidden_dropout": 0.1,
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "is_encoder_decoder": true,
+  "layer_norm_eps": 1e-05,
+  "mask_feature_length": 10,
+  "mask_feature_min_masks": 0,
+  "mask_feature_prob": 0.0,
+  "mask_time_length": 10,
+  "mask_time_min_masks": 2,
+  "mask_time_prob": 0.05,
+  "max_length": 1876,
+  "max_speech_positions": 1876,
+  "max_text_positions": 600,
+  "model_type": "speecht5",
+  "num_conv_pos_embedding_groups": 16,
+  "num_conv_pos_embeddings": 128,
+  "num_feat_extract_layers": 7,
+  "num_mel_bins": 80,
+  "pad_token_id": 1,
+  "positional_dropout": 0.1,
+  "reduction_factor": 2,
+  "scale_embedding": false,
+  "speaker_embedding_dim": 512,
+  "speech_decoder_postnet_dropout": 0.5,
+  "speech_decoder_postnet_kernel": 5,
+  "speech_decoder_postnet_layers": 5,
+  "speech_decoder_postnet_units": 256,
+  "speech_decoder_prenet_dropout": 0.5,
+  "speech_decoder_prenet_layers": 2,
+  "speech_decoder_prenet_units": 256,
+  "torch_dtype": "float32",
+  "transformers_version": "4.28.1",
+  "use_cache": true,
+  "use_guided_attention_loss": true,
+  "vocab_size": 81
+}

preprocessor_config.json

@@ -0,0 +1,23 @@
+{
+  "do_normalize": false,
+  "feature_extractor_type": "SpeechT5FeatureExtractor",
+  "feature_size": 1,
+  "fmax": 7600,
+  "fmin": 80,
+  "frame_signal_scale": 1.0,
+  "hop_length": 16,
+  "mel_floor": 1e-10,
+  "n_fft": 1024,
+  "n_freqs": 513,
+  "num_mel_bins": 80,
+  "padding_side": "right",
+  "padding_value": 0.0,
+  "processor_class": "SpeechT5Processor",
+  "reduction_factor": 2,
+  "return_attention_mask": true,
+  "sample_size": 1024,
+  "sample_stride": 256,
+  "sampling_rate": 16000,
+  "win_function": "hann_window",
+  "win_length": 64
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eccce95745831a1915292135e542097cc4d6bf9cb966331c1e99d6da40aa2330
+size 585479749

special_tokens_map.json

@@ -0,0 +1,13 @@
+{
+  "bos_token": "<s>",
+  "eos_token": "</s>",
+  "mask_token": {
+    "content": "<mask>",
+    "lstrip": true,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "<pad>",
+  "unk_token": "<unk>"
+}

spm_char.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7fcc48f3e225f627b1641db410ceb0c8649bd2b0c982e150b03f8be3728ab560
+size 238473

tokenizer_config.json

@@ -0,0 +1,11 @@
+{
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": true,
+  "eos_token": "</s>",
+  "model_max_length": 600,
+  "pad_token": "<pad>",
+  "processor_class": "SpeechT5Processor",
+  "sp_model_kwargs": {},
+  "tokenizer_class": "SpeechT5Tokenizer",
+  "unk_token": "<unk>"
+}

tts.py

@@ -0,0 +1,47 @@
+import torch
+from datasets import load_dataset
+from transformers import Pipeline, SpeechT5Processor, SpeechT5HifiGan
+
+
+class TTSPipeline(Pipeline):
+    def __init__(self, *args, vocoder=None, processor=None, **kwargs):
+        super().__init__(*args, **kwargs)
+
+        if vocoder is None:
+            raise ValueError("Must pass a vocoder to the TTSPipeline.")
+
+        if processor is None:
+            raise ValueError("Must pass a processor to the TTSPipeline.")
+
+        if isinstance(vocoder, str):
+            vocoder = SpeechT5HifiGan.from_pretrained(vocoder)
+
+        if isinstance(processor, str):
+            processor = SpeechT5Processor.from_pretrained(processor)
+
+        self.processor = processor
+        self.vocoder = vocoder
+
+    def preprocess(self, text, speaker_embeddings=None):
+        inputs = self.processor(text=text, return_tensors='pt')
+
+        if speaker_embeddings is None:
+            embeddings_dataset = load_dataset("Matthijs/cmu-arctic-xvectors", split="validation")
+            speaker_embeddings = torch.tensor(embeddings_dataset[7305]["xvector"]).unsqueeze(0)
+
+        return {'inputs': inputs, 'speaker_embeddings': speaker_embeddings}
+
+    def _forward(self, model_inputs):
+        inputs = model_inputs['inputs']
+        speaker_embeddings = model_inputs['speaker_embeddings']
+
+        with torch.no_grad():
+            speech = self.model.generate_speech(inputs['input_ids'], speaker_embeddings, vocoder=self.vocoder)
+
+        return speech
+
+    def _sanitize_parameters(self, **pipeline_parameters):
+        return {}, {}, {}
+
+    def postprocess(self, speech):
+        return speech