voxtream / app.py
herimor's picture
Add download button
413c2da
import os
import uuid
# Disable PyTorch dynamo/inductor globally
os.environ["TORCHDYNAMO_DISABLE"] = "1"
os.environ["TORCHINDUCTOR_DISABLE"] = "1"
import torch._dynamo as dynamo
dynamo.config.suppress_errors = True
import json
from pathlib import Path
import nltk
import torch
import spaces
import gradio as gr
import numpy as np
import soundfile as sf
from voxtream.generator import SpeechGenerator, SpeechGeneratorConfig
with open("configs/generator.json") as f:
config = SpeechGeneratorConfig(**json.load(f))
# Loading speaker encoder
torch.hub.load(
config.spk_enc_repo,
config.spk_enc_model,
model_name=config.spk_enc_model_name,
train_type=config.spk_enc_train_type,
dataset=config.spk_enc_dataset,
trust_repo=True,
verbose=False,
)
# Loading NLTK packages
nltk.download("averaged_perceptron_tagger_eng", quiet=True, raise_on_error=True)
nltk.download("punkt", quiet=True, raise_on_error=True)
# Initialize speech generator
speech_generator = SpeechGenerator(config)
FADE_OUT_SEC = 0.10
MIN_CHUNK_SEC = 0.2
CHUNK_SIZE = int(config.mimi_sr * MIN_CHUNK_SEC)
CUSTOM_CSS = """
/* overall width */
.gradio-container {max-width: 1100px !important}
/* stack labels tighter and even heights */
#cols .wrap > .form {gap: 10px}
#left-col, #right-col {gap: 14px}
/* make submit centered + bigger */
#submit {width: 260px; margin: 10px auto 0 auto;}
/* make clear align left and look secondary */
#clear {width: 120px;}
/* give audio a little breathing room */
audio {outline: none;}
"""
def float32_to_int16(audio_float32: np.ndarray) -> np.ndarray:
"""
Convert float32 audio samples (-1.0 to 1.0) to int16 PCM samples.
Parameters:
audio_float32 (np.ndarray): Input float32 audio samples.
Returns:
np.ndarray: Output int16 audio samples.
"""
if audio_float32.dtype != np.float32:
raise ValueError("Input must be a float32 numpy array")
# Clip to avoid overflow after scaling
audio_clipped = np.clip(audio_float32, -1.0, 1.0)
# Scale and convert
audio_int16 = (audio_clipped * 32767).astype(np.int16)
return audio_int16
def _clear_outputs():
# clears the player + hides file (download btn mirrors file via .change)
return None, gr.update(value=None, visible=False)
@spaces.GPU
def synthesize_fn(prompt_audio_path, prompt_text, target_text):
if next(speech_generator.model.parameters()).device.type == "cpu":
speech_generator.model.to("cuda")
speech_generator.mimi.to("cuda")
speech_generator.spk_enc.to("cuda")
speech_generator.aligner.aligner.to("cuda")
speech_generator.aligner.device = "cuda"
speech_generator.device = "cuda"
if not prompt_audio_path or not target_text:
return None, gr.update(value=None, visible=False)
stream = speech_generator.generate_stream(
prompt_text=prompt_text,
prompt_audio_path=Path(prompt_audio_path),
text=target_text,
)
buffer = []
buffer_len = 0
total_buffer = []
for frame, _ in stream:
buffer.append(frame)
total_buffer.append(frame)
buffer_len += frame.shape[0]
if buffer_len >= CHUNK_SIZE:
audio = np.concatenate(buffer)
yield (config.mimi_sr, float32_to_int16(audio)), None
# Reset buffer and length
buffer = []
buffer_len = 0
# Handle any remaining audio in the buffer
if buffer_len > 0:
final = np.concatenate(buffer)
nfade = min(int(config.mimi_sr * FADE_OUT_SEC), final.shape[0])
if nfade > 0:
fade = np.linspace(1.0, 0.0, nfade, dtype=np.float32)
final[-nfade:] *= fade
yield (config.mimi_sr, float32_to_int16(final)), None
# Save the full audio to a file for download
if len(total_buffer) > 0:
full_audio = np.concatenate(total_buffer)
nfade = min(int(config.mimi_sr * FADE_OUT_SEC), full_audio.shape[0])
if nfade > 0:
fade = np.linspace(1.0, 0.0, nfade, dtype=np.float32)
full_audio[-nfade:] *= fade
file_path = f"/tmp/voxtream_{uuid.uuid4().hex}.wav"
sf.write(file_path, float32_to_int16(full_audio), config.mimi_sr)
yield None, gr.update(value=file_path, visible=True)
else:
yield None, gr.update(value=None, visible=False)
def main():
with gr.Blocks(css=CUSTOM_CSS, title="VoXtream") as demo:
gr.Markdown("# VoXtream TTS demo")
gr.Markdown("⚠️ The initial latency can be high due to deployment on ZeroGPU. For faster inference, please try local deployment. For more details, please visit [VoXtream GitHub repo](https://github.com/herimor/voxtream)")
with gr.Row(equal_height=True, elem_id="cols"):
with gr.Column(scale=1, elem_id="left-col"):
prompt_audio = gr.Audio(
sources=["microphone", "upload"],
type="filepath",
label="Prompt audio (3-5 sec of target voice. Max 10 sec)",
)
prompt_text = gr.Textbox(
lines=3,
max_length=config.max_prompt_chars,
label=f"Prompt transcript (Required, max {config.max_prompt_chars} chars)",
placeholder="Text that matches the prompt audio",
)
with gr.Column(scale=1, elem_id="right-col"):
target_text = gr.Textbox(
lines=3,
max_length=config.max_phone_tokens,
label=f"Target text (Required, max {config.max_phone_tokens} chars)",
placeholder="What you want the model to say",
)
output_audio = gr.Audio(
label="Synthesized audio",
interactive=False,
streaming=True,
autoplay=True,
show_download_button=False,
show_share_button=False,
)
# appears only when file is ready
download_btn = gr.DownloadButton(
"Download audio",
visible=False,
)
with gr.Row():
clear_btn = gr.Button("Clear", elem_id="clear", variant="secondary")
submit_btn = gr.Button("Submit", elem_id="submit", variant="primary")
# Message box for validation errors
validation_msg = gr.Markdown("", visible=False)
# --- Validation logic ---
def validate_inputs(audio, ptext, ttext):
if not audio:
return gr.update(visible=True, value="⚠️ Please provide a prompt audio."), gr.update(interactive=False)
if not ptext.strip():
return gr.update(visible=True, value="⚠️ Please provide a prompt transcript."), gr.update(interactive=False)
if not ttext.strip():
return gr.update(visible=True, value="⚠️ Please provide target text."), gr.update(interactive=False)
return gr.update(visible=False, value=""), gr.update(interactive=True)
# Live validation whenever inputs change
for inp in [prompt_audio, prompt_text, target_text]:
inp.change(
fn=validate_inputs,
inputs=[prompt_audio, prompt_text, target_text],
outputs=[validation_msg, submit_btn],
)
# clear outputs before streaming
submit_btn.click(
fn=lambda a, p, t: (None, gr.update(value=None, visible=False)),
inputs=[prompt_audio, prompt_text, target_text],
outputs=[output_audio, download_btn],
show_progress="hidden",
).then(
fn=synthesize_fn,
inputs=[prompt_audio, prompt_text, target_text],
outputs=[output_audio, download_btn],
)
clear_btn.click(
fn=lambda: (
None, "", "", # inputs
None, # output_audio
gr.update(value=None, visible=False), # download_btn
gr.update(visible=False, value=""), # validation_msg
gr.update(interactive=False), # submit_btn
),
inputs=[],
outputs=[prompt_audio, prompt_text, target_text, output_audio, download_btn, validation_msg, submit_btn],
)
# --- Add Examples ---
gr.Markdown("### Examples")
ex = gr.Examples(
examples=[
[
"assets/app/male.wav",
"You could take the easy route or a situation that makes sense which a lot of you do",
"Hey, how are you doing? I just uhm want to make sure everything is okay."
],
[
"assets/app/female.wav",
"I would certainly anticipate some pushback whereas most people know if you followed my work.",
"Hello, hello. Let's have a quick chat, uh, in an hour. I need to share something with you."
],
],
inputs=[prompt_audio, prompt_text, target_text],
outputs=[output_audio, download_btn],
fn=synthesize_fn,
cache_examples=False,
)
ex.dataset.click(
fn=_clear_outputs,
inputs=[],
outputs=[output_audio, download_btn],
queue=False,
)
demo.launch()
if __name__ == "__main__":
main()