Spaces:
Running
on
Zero
Running
on
Zero
File size: 3,535 Bytes
0ebcf15 20636e5 0ebcf15 23208c6 0ebcf15 23208c6 0ebcf15 23208c6 0ebcf15 23208c6 0ebcf15 23208c6 0ebcf15 23208c6 0ebcf15 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 |
import gradio as gr
import torch.cuda
from InferenceInterfaces.ControllableInterface import ControllableInterface
from Utility.utils import float2pcm
class TTSWebUI:
def __init__(self, gpu_id="cpu", title="Controllable Text-to-Speech with IMS Toucan", article="", available_artificial_voices=1000):
self.controllable_ui = ControllableInterface(gpu_id=gpu_id,
available_artificial_voices=available_artificial_voices)
self.iface = gr.Interface(fn=self.read,
inputs=[gr.Textbox(lines=2,
placeholder="write what you want the synthesis to read here...",
value="What I cannot create, I do not understand.",
label="Text input"),
gr.Audio(type="filepath", show_label=True, container=True, label="Voice to Clone (if left empty, will use an artificial voice instead)"),
gr.Slider(minimum=0.0, maximum=0.8, step=0.1, value=0.4, label="Prosody Creativity"),
gr.Slider(minimum=0.7, maximum=1.3, step=0.1, value=1.0, label="Duration Scale"),
gr.Slider(minimum=0, maximum=available_artificial_voices, step=1,
value=279,
label="Random Seed for the artificial Voice"),
gr.Slider(minimum=-10.0, maximum=10.0, step=0.1, value=0.0, label="Femininity / Masculinity of artificial Voice"),
gr.Slider(minimum=-10.0, maximum=10.0, step=0.1, value=0.0, label="Voice Depth of artificial Voice")
],
outputs=[gr.Audio(type="numpy", label="Speech"),
gr.Image(label="Visualization")],
title=title,
theme="default",
allow_flagging="never",
article=article)
self.iface.launch()
def read(self,
prompt,
audio,
prosody_creativity,
duration_scaling_factor,
voice_seed,
emb1,
emb2
):
sr, wav, fig = self.controllable_ui.read(prompt,
audio,
voice_seed,
prosody_creativity,
duration_scaling_factor,
1.,
1.,
1.,
emb1,
emb2,
0.,
0.,
0.,
0.,
-24.)
return (sr, float2pcm(wav)), fig
if __name__ == '__main__':
TTSWebUI(gpu_id="cuda" if torch.cuda.is_available() else "cpu")
|