+
+| | |
+| ------------------------------- | --------------------------------------- |
+| 🐸💬 **CoquiTTS** | |
+| 💼 **Documentation** | [ReadTheDocs](https://tts.readthedocs.io/en/latest/)
+| 👩💻 **Questions** | [GitHub Discussions](https://github.com/coqui-ai/TTS/discussions) |
+| 🗯 **Community** | [](https://discord.gg/5eXr5seRrv) |
+
+
+"""
+
+article = """
+By using this demo you agree to the terms of the Coqui Public Model License at https://coqui.ai/cpml
+We collect data only for error cases for improvement.
+
+ """
+ )
+ with gr.Column():
+ # placeholder to align the image
+ pass
+
+ with gr.Row():
+ with gr.Column():
+ gr.Markdown(description)
+ with gr.Column():
+ gr.Markdown(links)
+
+ with gr.Row():
+ with gr.Column():
+ input_text_gr = gr.Textbox(
+ label="Text Prompt",
+ info="One or two sentences at a time is better. Up to 200 text characters.",
+ value="Tere, olen sinu uus häälekloon. Ürita mulle lindistada võimalikult hea kvaliteediga klipp, et oskaksin su häält paremini jäljendada.",
+ )
+ language_gr = gr.Dropdown(
+ label="Language",
+ info="Select an output language for the synthesised speech",
+ choices=[
+ "en",
+ "es",
+ "et",
+ "fr",
+ "de",
+ "it",
+ "pt",
+ "pl",
+ "tr",
+ "ru",
+ "nl",
+ "cs",
+ "ar",
+ "zh-cn",
+ "ja",
+ "ko",
+ "hu",
+ "hi"
+ ],
+ max_choices=1,
+ value="et",
+ )
+ ref_gr = gr.Audio(
+ label="Reference Audio",
+ info="Click on the ✎ button to upload your own target speaker audio",
+ type="filepath",
+ value="examples/LJ001-0030.wav",
+ )
+ mic_gr = gr.Audio(
+ source="microphone",
+ type="filepath",
+ info="Use your microphone to record audio",
+ label="Use Microphone for Reference",
+ )
+ use_mic_gr = gr.Checkbox(
+ label="Use Microphone",
+ value=False,
+ info="Notice: Microphone input may not work properly under traffic",
+ )
+ clean_ref_gr = gr.Checkbox(
+ label="Cleanup Reference Voice",
+ value=False,
+ info="This check can improve output if your microphone or reference voice is noisy",
+ )
+ auto_det_lang_gr = gr.Checkbox(
+ label="Do not use language auto-detect",
+ value=False,
+ info="Check to disable language auto-detection",
+ )
+ tos_gr = gr.Checkbox(
+ label="Agree",
+ value=False,
+ info="I agree to the terms of the CPML: https://coqui.ai/cpml",
+ )
+
+ tts_button = gr.Button("Send", elem_id="send-btn", visible=True)
+
+
+ with gr.Column():
+ video_gr = gr.Video(label="Waveform Visual")
+ audio_gr = gr.Audio(label="Synthesised Audio", autoplay=True)
+ out_text_gr = gr.Text(label="Metrics")
+ ref_audio_gr = gr.Audio(label="Reference Audio Used")
+
+ with gr.Row():
+ gr.Examples(examples,
+ label="Examples",
+ inputs=[input_text_gr, language_gr, ref_gr, mic_gr, use_mic_gr, clean_ref_gr, auto_det_lang_gr, tos_gr],
+ outputs=[video_gr, audio_gr, out_text_gr, ref_audio_gr],
+ fn=predict,
+ cache_examples=False,)
+
+ tts_button.click(predict, [input_text_gr, language_gr, ref_gr, mic_gr, use_mic_gr, clean_ref_gr, auto_det_lang_gr, tos_gr], outputs=[video_gr, audio_gr, out_text_gr, ref_audio_gr])
if __name__ == "__main__":
- demo.launch()
+ demo.queue()
+ demo.launch(debug=True, show_api=True)
\ No newline at end of file
diff --git a/app_local.py b/app_local.py
index 36476f81d1e64990d6d8df05a39c15a7e0745547..b650723155bf91fedb9fc4baefcdd2f3d772fee5 100644
--- a/app_local.py
+++ b/app_local.py
@@ -10,11 +10,18 @@ def load_model():
config = XttsConfig()
config.load_json("model/config.json")
XTTS_MODEL = Xtts.init_from_config(config)
- XTTS_MODEL.load_checkpoint(config, checkpoint_path="model/model.pth", vocab_path="model/vocab.json", use_deepspeed=False)
+ XTTS_MODEL.load_checkpoint(
+ config,
+ checkpoint_path="model/model.pth",
+ vocab_path="model/vocab.json",
+ eval=True,
+ use_deepspeed=False
+ )
XTTS_MODEL.to(device)
return XTTS_MODEL
model = load_model()
+#model.cuda()
def predict(sentence, language, reference_clip):
if not reference_clip or not reference_clip.split('.')[-1] in ['mp3', 'wav']:
@@ -32,11 +39,11 @@ def predict(sentence, language, reference_clip):
language=language,
gpt_cond_latent=gpt_cond_latent,
speaker_embedding=speaker_embedding,
- temperature=0.1,
- length_penalty=1.0,
- repetition_penalty=10.0,
- top_k=10,
- top_p=0.3,
+ temperature=model.config.temperature,
+ length_penalty=model.config.length_penalty,
+ repetition_penalty=model.config.repetition_penalty,
+ top_k=model.config.top_k,
+ top_p=model.config.top_p,
):
if chunk is not None:
wav_chunks.append(chunk)
@@ -52,4 +59,5 @@ demo = gr.Interface(
)
if __name__ == "__main__":
+ demo.queue()
demo.launch()
diff --git a/examples/.DS_Store b/examples/.DS_Store
new file mode 100644
index 0000000000000000000000000000000000000000..c9cfefd2c143f6c287f40306e7362fae18b01285
Binary files /dev/null and b/examples/.DS_Store differ
diff --git a/examples/LJ001-0030.wav b/examples/LJ001-0030.wav
new file mode 100644
index 0000000000000000000000000000000000000000..a1dc618bef0cbc2f3b38bf1c61d684f976804567
--- /dev/null
+++ b/examples/LJ001-0030.wav
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:62f11c8b1a6e728f4505bf32b03cf1d2afdad56cfd9893b6c6410409ec76b285
+size 304998
diff --git a/packages.txt b/packages.txt
new file mode 100644
index 0000000000000000000000000000000000000000..a9f1eea092d5e971b5475b82ee835cec7f196bad
--- /dev/null
+++ b/packages.txt
@@ -0,0 +1 @@
+ffmpeg
\ No newline at end of file
diff --git a/requirements.txt b/requirements.txt
index 44d0dd0df11ff51d272f91d524d226a0e088ed20..db849df4aac200176adac1199998921e31b38fb8 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -6,16 +6,16 @@ torch>=2.1
torchaudio>2.1.0
soundfile>=0.12.0
librosa>=0.10.1
-###scikit-learn>=1.3.0
+scikit-learn>=1.3.0
numba>0.58.0
inflect>=5.6.0
tqdm>=4.64.1
anyascii>=0.3.0
pyyaml>=6.0
-fsspec[http]>=2023.6.0
+fsspec>=2023.6.0
aiohttp>=3.8.1
packaging>=23.1
-###mutagen==1.47.0
+mutagen==1.47.0
# deps for examples
flask>=3.0.0
# deps for inference
@@ -48,21 +48,29 @@ bangla>=0.0.2
bnnumerizer>=0.0.2
bnunicodenormalizer>=0.1.0
# deps for japanese
-mecab-python3>=1.0.2
+mecab-python3>=1.0.6
unidic-lite==1.0.8
+unidic==1.1.0
cutlet>=0.2.0
# deps for estonian
tts_preprocess_et @ git+https://github.com/TartuNLP/tts_preprocess_et.git@v1.0.1
-#deps for tortoise
+# deps for tortoise
einops>=0.6.0
transformers>=4.43.0,<=4.46.2
#deps for bark
encodec>=0.1.1
# deps for XTTS
-###unidecode>=1.3.2
+unidecode>=1.3.2
num2words>=0.5.14
spacy[ja]>=3,<3.8
tokenizers==0.20.1
#deps for gradio
gradio
-huggingface_hub
\ No newline at end of file
+huggingface_hub
+pydantic==1.10.13
+python-multipart==0.0.6
+typing-extensions>=4.8.0
+langid
+deepspeed==0.14.5
+pydub
+py-cpuinfo
\ No newline at end of file