File size: 4,562 Bytes
635f007
 
73dbaa9
3851704
 
635f007
 
 
e2e4977
 
0675d4f
 
6eb9ea3
 
3851704
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b21342c
e27b102
 
3851704
e27b102
 
 
 
 
 
c00de09
635f007
3851704
d430de8
 
 
39e26fe
c6b1398
d430de8
 
 
4b8ade9
635f007
3851704
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
import gradio as gr
import styletts2importable
import numpy as np
import re

theme = gr.themes.Base(
    font=[gr.themes.GoogleFont('Libre Franklin'), gr.themes.GoogleFont('Public Sans'), 'system-ui', 'sans-serif'],
)
voicelist = ['f-us-1', 'f-us-2', 'f-us-3', 'f-us-4', 'm-us-1', 'm-us-2', 'm-us-3', 'm-us-4']
voices = {}
import phonemizer
global_phonemizer = phonemizer.backend.EspeakBackend(language='en-us', preserve_punctuation=True,  with_stress=True)
for v in voicelist:
    voices[v] = styletts2importable.compute_style(f'voices/{v}.wav')
def split_and_recombine_text(text, desired_length=200, max_length=400):
    """Split text it into chunks of a desired length trying to keep sentences intact."""
    # normalize text, remove redundant whitespace and convert non-ascii quotes to ascii
    text = re.sub(r'\n\n+', '\n', text)
    text = re.sub(r'\s+', ' ', text)
    text = re.sub(r'[“”]', '"', text)

    rv = []
    in_quote = False
    current = ""
    split_pos = []
    pos = -1
    end_pos = len(text) - 1

    def seek(delta):
        nonlocal pos, in_quote, current
        is_neg = delta < 0
        for _ in range(abs(delta)):
            if is_neg:
                pos -= 1
                current = current[:-1]
            else:
                pos += 1
                current += text[pos]
            if text[pos] == '"':
                in_quote = not in_quote
        return text[pos]

    def peek(delta):
        p = pos + delta
        return text[p] if p < end_pos and p >= 0 else ""

    def commit():
        nonlocal rv, current, split_pos
        rv.append(current)
        current = ""
        split_pos = []

    while pos < end_pos:
        c = seek(1)
        # do we need to force a split?
        if len(current) >= max_length:
            if len(split_pos) > 0 and len(current) > (desired_length / 2):
                # we have at least one sentence and we are over half the desired length, seek back to the last split
                d = pos - split_pos[-1]
                seek(-d)
            else:
                # no full sentences, seek back until we are not in the middle of a word and split there
                while c not in '!?.\n ' and pos > 0 and len(current) > desired_length:
                    c = seek(-1)
            commit()
        # check for sentence boundaries
        elif not in_quote and (c in '!?\n' or (c == '.' and peek(1) in '\n ')):
            # seek forward if we have consecutive boundary markers but still within the max length
            while pos < len(text) - 1 and len(current) < max_length and peek(1) in '!?.':
                c = seek(1)
            split_pos.append(pos)
            if len(current) >= desired_length:
                commit()
        # treat end of quote as a boundary if its followed by a space or newline
        elif in_quote and peek(1) == '"' and peek(2) in '\n ':
            seek(2)
            split_pos.append(pos)
    rv.append(current)

    # clean up, remove lines with only whitespace or punctuation
    rv = [s.strip() for s in rv]
    rv = [s for s in rv if len(s) > 0 and not re.match(r'^[\s\.,;:!?]*$', s)]

    return rv

def synthesize(text, voice, lngsteps, password, progress=gr.Progress()):
    if text.strip() == "":
        raise gr.Error("You must enter some text")
        
    texts = split_and_recombine_text(text)
    v = voice.lower()
    audios = []
    for t in progress.tqdm(texts):
        audios.append(styletts2importable.inference(t, voices[v], alpha=0.3, beta=0.7, diffusion_steps=lngsteps, embedding_scale=1))
    return (24000, np.concatenate(audios))


with gr.Blocks() as demo:
    with gr.Row():
        with gr.Column(scale=1):
            inp = gr.Textbox(label="Text", info="What would you like StyleTTS 2 to read? It works better on full sentences.", interactive=True)
            voice = gr.Dropdown(voicelist, label="Voice", info="Select a default voice.", value='m-us-2', interactive=True)
            multispeakersteps = gr.Slider(minimum=3, maximum=15, value=3, step=1, label="Diffusion Steps", info="Theoretically, higher should be better quality but slower, but we cannot notice a difference. Try with lower steps first - it is faster", interactive=True)
        with gr.Column(scale=1):
            btn = gr.Button("Synthesize", variant="primary")
            audio = gr.Audio(interactive=False, label="Synthesized Audio")
            btn.click(synthesize, inputs=[inp, voice, multispeakersteps], outputs=[audio], concurrency_limit=4)

demo.queue(api_open=True, max_size=15).launch(show_api=True)