From 496a4be3e35e127b6df09033ff3b57a0a00c970e Mon Sep 17 00:00:00 2001 From: Edresson Date: Thu, 30 Jul 2020 03:51:20 -0300 Subject: [PATCH] add support for synthesize using variable size external embedding and add bugfix in scipy.io import --- mozilla_voice_tts/bin/train_tts.py | 1 - mozilla_voice_tts/tts/utils/synthesis.py | 3 --- 2 files changed, 4 deletions(-) diff --git a/mozilla_voice_tts/bin/train_tts.py b/mozilla_voice_tts/bin/train_tts.py index daa517b9..1b9bc032 100644 --- a/mozilla_voice_tts/bin/train_tts.py +++ b/mozilla_voice_tts/bin/train_tts.py @@ -523,7 +523,6 @@ def main(args): # pylint: disable=redefined-outer-name "a previously trained model." elif c.use_external_speaker_embedding_file and c.external_speaker_embedding_file: # if start new train using External Embedding file speaker_mapping = load_speaker_mapping(c.external_speaker_embedding_file) - print(speaker_mapping) speaker_embedding_dim = len(speaker_mapping[list(speaker_mapping.keys())[0]]['embedding']) elif c.use_external_speaker_embedding_file and not c.external_speaker_embedding_file: # if start new train using External Embedding file and don't pass external embedding file raise "use_external_speaker_embedding_file is True, so you need pass a external speaker embedding file, run GE2E-Speaker_Encoder-ExtractSpeakerEmbeddings-by-sample.ipynb or AngularPrototypical-Speaker_Encoder-ExtractSpeakerEmbeddings-by-sample.ipynb notebook in notebooks/ folder" diff --git a/mozilla_voice_tts/tts/utils/synthesis.py b/mozilla_voice_tts/tts/utils/synthesis.py index 52b33e86..2f746533 100644 --- a/mozilla_voice_tts/tts/utils/synthesis.py +++ b/mozilla_voice_tts/tts/utils/synthesis.py @@ -210,13 +210,10 @@ def synthesis(model, if backend == 'torch': if speaker_id is not None: speaker_id = id_to_torch(speaker_id, cuda=use_cuda) -<<<<<<< HEAD:mozilla_voice_tts/tts/utils/synthesis.py if speaker_embedding is not None: speaker_embedding = embedding_to_torch(speaker_embedding, cuda=use_cuda) -======= ->>>>>>> Added support for Tacotron2 GST + abbility to condition style input with wav or tokens:utils/synthesis.py if not isinstance(style_mel, dict): style_mel = numpy_to_torch(style_mel, torch.float, cuda=use_cuda) inputs = numpy_to_torch(inputs, torch.long, cuda=use_cuda)