Spaces:
Sleeping
Sleeping
Samuel L Meyers
commited on
Commit
·
da8a172
1
Parent(s):
e71462a
Inital MiniChat test
Browse files
app.py
CHANGED
@@ -1,44 +1,11 @@
|
|
1 |
-
"""
|
2 |
-
Copyright 2022 Balacoon
|
3 |
-
|
4 |
-
TTS interactive demo
|
5 |
-
"""
|
6 |
-
|
7 |
-
import os
|
8 |
-
import glob
|
9 |
import logging
|
10 |
from typing import cast
|
11 |
from threading import Lock
|
12 |
-
from transformers import
|
13 |
-
|
14 |
-
import gradio as gr
|
15 |
-
from balacoon_tts import TTS
|
16 |
-
from huggingface_hub import hf_hub_download, list_repo_files
|
17 |
import torch
|
18 |
|
19 |
from conversation import get_default_conv_template
|
20 |
-
|
21 |
-
# locker that disallow access to the tts object from more then one thread
|
22 |
-
locker = Lock()
|
23 |
-
# global tts module, initialized from a model selected
|
24 |
-
tts = None
|
25 |
-
# path to the model that is currently used in tts
|
26 |
-
cur_model_path = None
|
27 |
-
# cache of speakers, maps model name to speaker list
|
28 |
-
model_to_speakers = dict()
|
29 |
-
model_repo_dir = "/data"
|
30 |
-
for name in list_repo_files(repo_id="balacoon/tts"):
|
31 |
-
if not os.path.isfile(os.path.join(model_repo_dir, name)):
|
32 |
-
hf_hub_download(
|
33 |
-
repo_id="balacoon/tts",
|
34 |
-
filename=name,
|
35 |
-
local_dir=model_repo_dir,
|
36 |
-
)
|
37 |
-
|
38 |
-
stt_pipe = pipeline(
|
39 |
-
task="automatic-speech-recognition",
|
40 |
-
model="openai/whisper-large-v3",
|
41 |
-
)
|
42 |
|
43 |
talkers = {
|
44 |
"m3b": {
|
@@ -48,12 +15,6 @@ talkers = {
|
|
48 |
}
|
49 |
}
|
50 |
|
51 |
-
def transcribe_stt(audio):
|
52 |
-
if audio is None:
|
53 |
-
raise gr.Error("No audio file submitted! Please upload or record an audio file before submitting your request.")
|
54 |
-
text = stt_pipe(audio, generate_kwargs={"language": "english", "task": "transcribe"})["text"]
|
55 |
-
return text
|
56 |
-
|
57 |
def m3b_talk(text):
|
58 |
m3bconv = talkers["m3b"]["conv"]
|
59 |
m3bconv.append_message(m3bconv.roles[0], text)
|
@@ -73,75 +34,6 @@ def main():
|
|
73 |
logging.basicConfig(level=logging.INFO)
|
74 |
|
75 |
with gr.Blocks() as demo:
|
76 |
-
gr.Markdown(
|
77 |
-
"""
|
78 |
-
<h1 align="center">Balacoon🦝 Text-to-Speech</h1>
|
79 |
-
|
80 |
-
1. Write an utterance to generate,
|
81 |
-
2. Select the model to synthesize with
|
82 |
-
3. Select speaker
|
83 |
-
4. Hit "Generate" and listen to the result!
|
84 |
-
|
85 |
-
You can learn more about models available
|
86 |
-
[here](https://huggingface.co/balacoon/tts).
|
87 |
-
Visit [Balacoon website](https://balacoon.com/) for more info.
|
88 |
-
"""
|
89 |
-
)
|
90 |
-
with gr.Row(variant="panel"):
|
91 |
-
text = gr.Textbox(label="Text", placeholder="Type something here...")
|
92 |
-
|
93 |
-
with gr.Row():
|
94 |
-
with gr.Column(variant="panel"):
|
95 |
-
repo_files = os.listdir(model_repo_dir)
|
96 |
-
model_files = [x for x in repo_files if x.endswith("_cpu.addon")]
|
97 |
-
model_name = gr.Dropdown(
|
98 |
-
label="Model",
|
99 |
-
choices=model_files,
|
100 |
-
)
|
101 |
-
with gr.Column(variant="panel"):
|
102 |
-
speaker = gr.Dropdown(label="Speaker", choices=[])
|
103 |
-
|
104 |
-
def set_model(model_name_str: str):
|
105 |
-
"""
|
106 |
-
gets value from `model_name`. either
|
107 |
-
uses cached list of speakers for the given model name
|
108 |
-
or loads the addon and checks what are the speakers.
|
109 |
-
"""
|
110 |
-
global model_to_speakers
|
111 |
-
if model_name_str in model_to_speakers:
|
112 |
-
speakers = model_to_speakers[model_name_str]
|
113 |
-
else:
|
114 |
-
global tts, cur_model_path, locker
|
115 |
-
with locker:
|
116 |
-
# need to load this model to learn the list of speakers
|
117 |
-
model_path = os.path.join(model_repo_dir, model_name_str)
|
118 |
-
if tts is not None:
|
119 |
-
del tts
|
120 |
-
tts = TTS(model_path)
|
121 |
-
cur_model_path = model_path
|
122 |
-
speakers = tts.get_speakers()
|
123 |
-
model_to_speakers[model_name_str] = speakers
|
124 |
-
|
125 |
-
value = speakers[-1]
|
126 |
-
return gr.Dropdown.update(
|
127 |
-
choices=speakers, value=value, visible=True
|
128 |
-
)
|
129 |
-
|
130 |
-
model_name.change(set_model, inputs=model_name, outputs=speaker)
|
131 |
-
|
132 |
-
with gr.Row(variant="panel"):
|
133 |
-
generate = gr.Button("Generate")
|
134 |
-
with gr.Row(variant="panel"):
|
135 |
-
audio = gr.Audio()
|
136 |
-
with gr.Row(variant="panel"):
|
137 |
-
gr.Markdown("## Transcribe\n\nTranscribe audio to text.")
|
138 |
-
with gr.Row(variant="panel"):
|
139 |
-
with gr.Column(variant="panel"):
|
140 |
-
stt_input_mic = gr.Audio(source="microphone", type="filepath", label="Record")
|
141 |
-
stt_input_file = gr.Audio(source="upload", type="filepath", label="Upload")
|
142 |
-
with gr.Column(variant="panel"):
|
143 |
-
stt_transcribe_output = gr.Textbox()
|
144 |
-
stt_transcribe_btn = gr.Button("Transcribe")
|
145 |
with gr.Row(variant="panel"):
|
146 |
gr.Markdown("## Talk to MiniChat-3B\n\nTalk to MiniChat-3B.")
|
147 |
with gr.Row(variant="panel"):
|
@@ -151,33 +43,6 @@ def main():
|
|
151 |
m3b_talk_output = gr.Textbox()
|
152 |
m3b_talk_btn = gr.Button("Send")
|
153 |
|
154 |
-
def synthesize_audio(text_str: str, model_name_str: str, speaker_str: str):
|
155 |
-
"""
|
156 |
-
gets utterance to synthesize from `text` Textbox
|
157 |
-
and speaker name from `speaker` dropdown list.
|
158 |
-
speaker name might be empty for single-speaker models.
|
159 |
-
Synthesizes the waveform and updates `audio` with it.
|
160 |
-
"""
|
161 |
-
if not text_str or not model_name_str or not speaker_str:
|
162 |
-
logging.info("text, model name or speaker are not provided")
|
163 |
-
return None
|
164 |
-
expected_model_path = os.path.join(model_repo_dir, model_name_str)
|
165 |
-
global tts, cur_model_path, locker
|
166 |
-
with locker:
|
167 |
-
if expected_model_path != cur_model_path:
|
168 |
-
# reload model
|
169 |
-
if tts is not None:
|
170 |
-
del tts
|
171 |
-
tts = TTS(expected_model_path)
|
172 |
-
cur_model_path = expected_model_path
|
173 |
-
if len(text_str) > 1024:
|
174 |
-
# truncate the text
|
175 |
-
text_str = text_str[:1024]
|
176 |
-
samples = tts.synthesize(text_str, speaker_str)
|
177 |
-
return gr.Audio.update(value=(tts.get_sampling_rate(), samples))
|
178 |
-
|
179 |
-
generate.click(synthesize_audio, inputs=[text, model_name, speaker], outputs=audio, api_name="synthesize")
|
180 |
-
stt_transcribe_btn.click(transcribe_stt, inputs=stt_input_file, outputs=stt_transcribe_output, api_name="transcribe")
|
181 |
m3b_talk_btn.click(m3b_talk, inputs=m3b_talk_input, outputs=m3b_talk_output, api_name="talk_m3b")
|
182 |
|
183 |
demo.queue(concurrency_count=1).launch()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
import logging
|
2 |
from typing import cast
|
3 |
from threading import Lock
|
4 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer
|
|
|
|
|
|
|
|
|
5 |
import torch
|
6 |
|
7 |
from conversation import get_default_conv_template
|
8 |
+
import gradio as gr
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
9 |
|
10 |
talkers = {
|
11 |
"m3b": {
|
|
|
15 |
}
|
16 |
}
|
17 |
|
|
|
|
|
|
|
|
|
|
|
|
|
18 |
def m3b_talk(text):
|
19 |
m3bconv = talkers["m3b"]["conv"]
|
20 |
m3bconv.append_message(m3bconv.roles[0], text)
|
|
|
34 |
logging.basicConfig(level=logging.INFO)
|
35 |
|
36 |
with gr.Blocks() as demo:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
37 |
with gr.Row(variant="panel"):
|
38 |
gr.Markdown("## Talk to MiniChat-3B\n\nTalk to MiniChat-3B.")
|
39 |
with gr.Row(variant="panel"):
|
|
|
43 |
m3b_talk_output = gr.Textbox()
|
44 |
m3b_talk_btn = gr.Button("Send")
|
45 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
46 |
m3b_talk_btn.click(m3b_talk, inputs=m3b_talk_input, outputs=m3b_talk_output, api_name="talk_m3b")
|
47 |
|
48 |
demo.queue(concurrency_count=1).launch()
|