Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Stream generated audio to browser in real-time #33

Open
wants to merge 2 commits into
base: main
Choose a base branch
from
Open

Stream generated audio to browser in real-time #33

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
82ca017
make audio stream in realtime as it is generated
MattJeanes Jun 16, 2024
d6eafe6
also include the full audio file at the end as before
MattJeanes Jun 16, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
29 changes: 21 additions & 8 deletions xtts_demo.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -57,14 +57,15 @@ def load_model(xtts_checkpoint, xtts_config, xtts_vocab,xtts_speaker):

def run_tts(lang, tts_text, speaker_audio_file, temperature, length_penalty,repetition_penalty,top_k,top_p,sentence_split,use_config):
if XTTS_MODEL is None or not speaker_audio_file:
return "You need to run the previous step to load the model !!", None, None
yield "You need to run the previous step to load the model !!", None, None, None
return



gpt_cond_latent, speaker_embedding = XTTS_MODEL.get_conditioning_latents(audio_path=speaker_audio_file, gpt_cond_len=XTTS_MODEL.config.gpt_cond_len, max_ref_length=XTTS_MODEL.config.max_ref_len, sound_norm_refs=XTTS_MODEL.config.sound_norm_refs)

if use_config:
out = XTTS_MODEL.inference(
generator = XTTS_MODEL.inference_stream(
text=tts_text,
language=lang,
gpt_cond_latent=gpt_cond_latent,
Expand All @@ -77,7 +78,7 @@ def run_tts(lang, tts_text, speaker_audio_file, temperature, length_penalty,repe
enable_text_splitting = True
)
else:
out = XTTS_MODEL.inference(
generator = XTTS_MODEL.inference_stream(
text=tts_text,
language=lang,
gpt_cond_latent=gpt_cond_latent,
Expand All @@ -90,13 +91,24 @@ def run_tts(lang, tts_text, speaker_audio_file, temperature, length_penalty,repe
enable_text_splitting = sentence_split
)

fs = 24000
duration_in_seconds = 0.1 # Duration of the silent audio chunk
silent_audio = np.zeros(int(fs * duration_in_seconds), dtype=np.float32)
yield "Starting..", (fs, silent_audio), None, speaker_audio_file

wavs = []
for output in generator:
wav = output.cpu()
wavs.append(wav)
yield "Generating..", (fs, torch.cat((wav,), dim=0).numpy()), None, speaker_audio_file

out_wav = torch.cat(wavs, dim=0).numpy()
with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as fp:
out["wav"] = torch.tensor(out["wav"]).unsqueeze(0)
out_wav = torch.tensor(out_wav).unsqueeze(0)
out_path = fp.name
torchaudio.save(out_path, out["wav"], 24000)

return "Speech generated !", out_path, speaker_audio_file
torchaudio.save(out_path, out_wav, fs)

yield "Finished!", None, out_path, speaker_audio_file

def load_params_tts(out_path,version):

Expand Down Expand Up @@ -577,6 +589,7 @@ def load_params(out_path):
progress_gen = gr.Label(
label="Progress:"
)
tts_audio_stream = gr.Audio(label="Live Audio Stream.", streaming=True)
tts_output_audio = gr.Audio(label="Generated Audio.")
reference_audio = gr.Audio(label="Reference audio used.")

Expand Down Expand Up @@ -662,7 +675,7 @@ def load_params(out_path):
sentence_split,
use_config
],
outputs=[progress_gen, tts_output_audio,reference_audio],
outputs=[progress_gen, tts_audio_stream, tts_output_audio, reference_audio],
)

load_params_tts_btn.click(
Expand Down