Spaces:
Runtime error
Runtime error
imseldrith
commited on
Commit
•
a07ed46
1
Parent(s):
eece182
Upload 3 files
Browse files- src/functions.py +53 -0
- src/main.py +101 -0
- src/schemas.py +9 -0
src/functions.py
ADDED
@@ -0,0 +1,53 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import time
|
2 |
+
import json
|
3 |
+
|
4 |
+
def get_time_utc(zone ,delay=0):
|
5 |
+
loc_time = time.gmtime(time.time() + delay + zone * 60 * 60)
|
6 |
+
return time.strftime("%Y-%m-%d %H:%M:%S",loc_time)
|
7 |
+
|
8 |
+
def clear_dict(d):
|
9 |
+
if d is None:
|
10 |
+
return None
|
11 |
+
elif isinstance(d, list):
|
12 |
+
return list(filter(lambda x: x is not None, map(clear_dict, d)))
|
13 |
+
elif not isinstance(d, dict):
|
14 |
+
return d
|
15 |
+
else:
|
16 |
+
r = dict(
|
17 |
+
filter(lambda x: x[1] is not None,
|
18 |
+
map(lambda x: (x[0], clear_dict(x[1])),
|
19 |
+
d.items())))
|
20 |
+
if not bool(r):
|
21 |
+
return None
|
22 |
+
return r
|
23 |
+
|
24 |
+
def print_env(server_port=6006, sleep=3):
|
25 |
+
print("")
|
26 |
+
print("")
|
27 |
+
print("###########################################")
|
28 |
+
print("environment variable start-----------------------------------")
|
29 |
+
print("###########################################")
|
30 |
+
print("")
|
31 |
+
|
32 |
+
print("server_port: " + str(server_port))
|
33 |
+
|
34 |
+
print("")
|
35 |
+
print("###########################################")
|
36 |
+
print("Please check the environment variables (the program will start in 3 seconds) ↑↑↑↑↑↑↑↑↑↑↑↑↑↑↑↑↑↑↑↑↑↑↑↑↑↑↑↑↑↑↑↑")
|
37 |
+
print("###########################################")
|
38 |
+
print("")
|
39 |
+
print("")
|
40 |
+
time.sleep(sleep)
|
41 |
+
return
|
42 |
+
|
43 |
+
def print_log(request, respose, time_start=0):
|
44 |
+
print("______________________________________________")
|
45 |
+
print("request" + ":::\n" + json.dumps(clear_dict(request.__dict__))) # class dict convert to json
|
46 |
+
print("respose" + ":::")
|
47 |
+
if isinstance(respose, dict) or isinstance(respose,list):
|
48 |
+
print(respose)
|
49 |
+
else:
|
50 |
+
print(respose.__dict__)
|
51 |
+
print("cost:::\n" + str(time.time() - time_start) + "s")
|
52 |
+
print("finish:::\n" + get_time_utc(-8))
|
53 |
+
return
|
src/main.py
ADDED
@@ -0,0 +1,101 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from fastapi import FastAPI
|
2 |
+
import schemas
|
3 |
+
import uvicorn
|
4 |
+
from starlette.middleware.cors import CORSMiddleware
|
5 |
+
from functions import *
|
6 |
+
import base64
|
7 |
+
import os
|
8 |
+
import traceback
|
9 |
+
|
10 |
+
from bark import SAMPLE_RATE, generate_audio, preload_models
|
11 |
+
import soundfile as sf
|
12 |
+
import wave
|
13 |
+
import numpy as np
|
14 |
+
import nltk
|
15 |
+
|
16 |
+
# fastapi port
|
17 |
+
server_port = 6006
|
18 |
+
|
19 |
+
# Preload model
|
20 |
+
preload_models()
|
21 |
+
|
22 |
+
app = FastAPI(docs_url=None, redoc_url=None)
|
23 |
+
|
24 |
+
# Set allowed access domain names
|
25 |
+
origins = ["*"] # set to "*" means all.
|
26 |
+
|
27 |
+
|
28 |
+
def concatenate_wavs(wav_files, output_file, silence_duration=0.3):
|
29 |
+
wavs = [wave.open(f, 'rb') for f in wav_files]
|
30 |
+
sampwidth = wavs[0].getsampwidth()
|
31 |
+
framerate = wavs[0].getframerate()
|
32 |
+
nchannels = wavs[0].getnchannels()
|
33 |
+
|
34 |
+
samples = [wav.readframes(wav.getnframes()) for wav in wavs]
|
35 |
+
total_frames = sum(len(s) for s in samples) + int(silence_duration * framerate * nchannels * sampwidth)
|
36 |
+
|
37 |
+
output = wave.open(output_file, 'wb')
|
38 |
+
output.setparams((nchannels, sampwidth, framerate, total_frames, 'NONE', 'Uncompressed'))
|
39 |
+
|
40 |
+
for s in samples:
|
41 |
+
output.writeframes(s)
|
42 |
+
silence_frame = np.zeros((int(silence_duration * framerate), 2)).astype(np.int16).tobytes()
|
43 |
+
for i in range(int(nchannels / 2)):
|
44 |
+
output.writeframes(silence_frame)
|
45 |
+
|
46 |
+
output.close()
|
47 |
+
|
48 |
+
|
49 |
+
# Set cross domain parameter transfer
|
50 |
+
app.add_middleware(
|
51 |
+
CORSMiddleware,
|
52 |
+
allow_origins=origins, # Set allowed origins sources
|
53 |
+
allow_credentials=True,
|
54 |
+
allow_methods=["*"], # Set up HTTP methods that allow cross domain access, such as get, post, put, etc.
|
55 |
+
allow_headers=["*"]) # Allowing cross domain headers can be used to identify sources and other functions.
|
56 |
+
|
57 |
+
|
58 |
+
@app.post("/tts_bark/")
|
59 |
+
async def tts_bark(item: schemas.generate_web):
|
60 |
+
time_start = time.time()
|
61 |
+
text = item.text
|
62 |
+
print(f"{text=}")
|
63 |
+
try:
|
64 |
+
sentences = nltk.sent_tokenize(text)
|
65 |
+
idx = 1
|
66 |
+
wavs = []
|
67 |
+
for s in sentences:
|
68 |
+
audio_array = generate_audio(s, history_prompt="en_speaker_8", text_temp=0.6, waveform_temp=0.6)
|
69 |
+
fname = f"tmp-{idx}.wav"
|
70 |
+
sf.write(fname, audio_array, SAMPLE_RATE)
|
71 |
+
idx += 1
|
72 |
+
wavs.append(fname)
|
73 |
+
file_name_pre = f"out-{time.time()}"
|
74 |
+
file_name_wav = file_name_pre + ".wav"
|
75 |
+
file_name_ogg = file_name_pre + ".ogg"
|
76 |
+
concatenate_wavs(wavs, file_name_wav)
|
77 |
+
|
78 |
+
# convert to OGG
|
79 |
+
os.system("ffmpeg -i " + file_name_wav + " -c:a libopus -b:a 64k -y " + file_name_ogg)
|
80 |
+
|
81 |
+
with open(file_name_ogg, "rb") as f:
|
82 |
+
audio_content = f.read()
|
83 |
+
base64_audio = base64.b64encode(audio_content).decode("utf-8")
|
84 |
+
res = {"file_base64": base64_audio,
|
85 |
+
"audio_text": text,
|
86 |
+
"file_name": file_name_ogg,
|
87 |
+
}
|
88 |
+
print_log(item, res, time_start)
|
89 |
+
os.remove(file_name_wav)
|
90 |
+
os.remove(file_name_ogg)
|
91 |
+
|
92 |
+
return res
|
93 |
+
except Exception as err:
|
94 |
+
res = {"code": 9, "msg": "api error", "err": str(err), "traceback": traceback.format_exc()}
|
95 |
+
print_log(item, res, time_start)
|
96 |
+
return res
|
97 |
+
|
98 |
+
if __name__ == '__main__':
|
99 |
+
|
100 |
+
print_env(server_port)
|
101 |
+
uvicorn.run(app="main:app", host="0.0.0.0", port=server_port, reload=False)
|
src/schemas.py
ADDED
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from pydantic import BaseModel
|
2 |
+
from typing import Union
|
3 |
+
|
4 |
+
|
5 |
+
class generate_web(BaseModel):
|
6 |
+
text: Union[str, None] = None
|
7 |
+
|
8 |
+
class Config:
|
9 |
+
orm_mode = True
|