File size: 5,027 Bytes
99f6290
 
 
 
 
34e0ba3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
---
license: apache-2.0
base_model:
- nvidia/NVLM-D-72B
---
!pip install -U scipy
!git clone https://github.com/neonbjb/tortoise-tts.git
%cd tortoise-tts
!pip install -r requirements.txt
!python setup.py install
!pip install gradio

import os
import gradio as gr
import torchaudio
import time
from datetime import datetime
from tortoise.api import TextToSpeech
from tortoise.utils.audio import load_audio, load_voice, load_voices
import os

# Set the Gradio queue flag to disabled
os.environ["COMMANDLINE_ARGS"] = "--no-gradio-queue"
VOICE_OPTIONS = [
    "random",  # special option for random voice
    "custom_voice",  # special option for custom voice
    "disabled",  # special option for disabled voice
]

def inference(text, emotion, prompt, voice, mic_audio, voice_b, voice_c, preset, seed):
    if voice != "custom_voice":
        voices = [voice]
    else:
        voices = []

    if voice_b != "disabled":
        voices.append(voice_b)
    if voice_c != "disabled":
        voices.append(voice_c)

    if emotion != "None/Custom":
        text = f"[I am really {emotion.lower()},] {text}"
    elif prompt.strip() != "":
        text = f"[{prompt},] {text}"

    c = None
    if voice == "custom_voice":
        if mic_audio is None:
            raise gr.Error("Please provide audio from mic when choosing custom voice")
        c = load_audio(mic_audio, 22050)

    if len(voices) == 1 or len(voices) == 0:
        if voice == "custom_voice":
            voice_samples, conditioning_latents = [c], None
        else:
            voice_samples, conditioning_latents = load_voice(voice)
    else:
        voice_samples, conditioning_latents = load_voices(voices)
        if voice == "custom_voice":
            voice_samples.extend([c])

    sample_voice = voice_samples[0] if len(voice_samples) else None

    start_time = time.time()
    gen, _ = tts.tts_with_preset(
        text,
        voice_samples=voice_samples,
        conditioning_latents=conditioning_latents,
        preset=preset,
        use_deterministic_seed=seed,
        return_deterministic_state=True,
        k=3,
    )

    with open("Tortoise_TTS_Runs.log", "a") as f:
        f.write(
            f"{datetime.now()} | Voice: {','.join(voices)} | Text: {text} | Quality: {preset} | Time Taken (s): {time.time()-start_time} | Seed: {seed}\n"
        )

    return (
        (22050, sample_voice.squeeze().cpu().numpy()),
        (24000, gen[0].squeeze().cpu().numpy()),
        (24000, gen[1].squeeze().cpu().numpy()),
        (24000, gen[2].squeeze().cpu().numpy()),
    )

def main():
    # Custom HTML for the title
    title_html = "<h1 style='text-align: center; color: orange; font-weight: bold;'>RJ VOICE CLONING</h1>"

    # Interface components
    text = gr.Textbox(lines=4, label="Text:")
    emotion = gr.Radio(
        ["None/Custom", "Happy", "Sad", "Angry", "Disgusted", "Arrogant"],
        value="None/Custom",
        label="Select emotion:",
        type="value",
    )
    prompt = gr.Textbox(lines=1, label="Enter prompt if [Custom] emotion:")
    preset = gr.Radio(
        ["ultra_fast", "fast", "standard", "high_quality"],
        value="fast",
        label="Preset mode (determines quality with tradeoff over speed):",
        type="value",
    )
    voice = gr.Dropdown(
        os.listdir(os.path.join("tortoise", "voices")) + VOICE_OPTIONS,
        value="angie",  # Default voice
        label="Select voice:",
        type="value",
    )
    mic_audio = gr.Audio(
        label="Record voice (when selected custom_voice):",
        type="filepath"
    )
    voice_b = gr.Dropdown(
        os.listdir(os.path.join("tortoise", "voices")) + VOICE_OPTIONS,
        value="disabled",
        label="(Optional) Select second voice:",
        type="value",
    )
    voice_c = gr.Dropdown(
        os.listdir(os.path.join("tortoise", "voices")) + VOICE_OPTIONS,
        value="disabled",
        label="(Optional) Select third voice:",
        type="value",
    )
    seed = gr.Number(value=0, precision=0, label="Seed (for reproducibility):")

    selected_voice = gr.Audio(label="Sample of selected voice (first):")
    output_audio_1 = gr.Audio(label="Output [Candidate 1]:")
    output_audio_2 = gr.Audio(label="Output [Candidate 2]:")
    output_audio_3 = gr.Audio(label="Output [Candidate 3]:")

    # Create the Gradio interface
    interface = gr.Interface(
        fn=inference,
        inputs=[text, emotion, prompt, voice, mic_audio, voice_b, voice_c, preset, seed],
        outputs=[selected_voice, output_audio_1, output_audio_2, output_audio_3],
        title="RJ VOICE CLONING",
        description=title_html,
        css=".gradio-container { background-color: black; color: orange; }"
    )

    # Launch the interface
    interface.launch(share=True)

if __name__ == "__main__":
    tts = TextToSpeech()

    with open("Tortoise_TTS_Runs.log", "a") as f:
        f.write(
            f"\n\n-------------------------Tortoise TTS Logs, {datetime.now()}-------------------------\n"
        )

    main()