fix

app.py

@@ -1,8 +1,9 @@
 import os
+import threading
 
 os.environ['TF_FORCE_GPU_ALLOW_GROWTH'] = 'true'
 
-from music.search import get_random_spit, get_albums
+from music.search import get_random_spit, get_albums, search_youtube, download_youtube, get_youtube, download_random
 from vits.models import SynthesizerInfer
 import whisper.inference
 from omegaconf import OmegaConf
@@ -17,7 +18,6 @@ import random
 from spleeter.separator import Separator
 from spleeter.audio.adapter import AudioAdapter
 from pydub import AudioSegment
-import scipy.io.wavfile
 import uuid
 
 import logging
@@ -178,14 +178,6 @@ def svc_change(argswave, argsspk):
     return out_audio
 
 
-def np_to_audio_segment(fp_arr):
-    wav_io = io.BytesIO()
-    scipy.io.wavfile.write(wav_io, 16000, fp_arr)
-    wav_io.seek(0)
-    sound = AudioSegment.from_wav(wav_io)
-    return sound
-
-
 def svc_main(sid, input_audio):
     if input_audio is None:
         return "You need to upload an audio", None
@@ -238,14 +230,26 @@ def auto_search(name):
     save_music_path = 'downloaded'
     if not os.path.exists(save_music_path):
         os.makedirs(save_music_path)
+
     config = {'logfilepath': 'musicdl.log', save_music_path: 'downloaded', 'search_size_per_source': 5, 'proxies': {}}
-    albums = get_albums(keywords=name, config=config)
-    if len(albums) == 0:
-        return "No music found", None
-    album = random.choice(albums)
-    save_path = get_random_spit(album)
-    fp = save_path
-    signal, sampling_rate = soundfile.read(fp)
+    save_path = os.path.join(save_music_path, name + '.mp3')
+    # youtube
+    task1 = threading.Thread(
+        target=get_youtube,
+        args=(name, save_path)
+    )
+    task1.start()
+    task2 = threading.Thread(
+        target=download_random,
+        args=(name, config, save_path)
+    )
+    task2.start()
+    task1.join()
+    task2.join()
+
+    if not os.path.exists(save_path):
+        return "Not Found", None
+    signal, sampling_rate = soundfile.read(save_path)
     return "Found a music", (sampling_rate, signal)
 
 

music/search.py

@@ -4,6 +4,11 @@ import random
 from musicdl import musicdl
 from musicdl.modules import Downloader
 from pydub import AudioSegment
+from yt_dlp import YoutubeDL
+import yt_dlp
+from yt_dlp.utils import download_range_func
+import json
+
 
 def is_integer(string):
     if string.isdigit():
@@ -46,6 +51,63 @@ def size_to_int(size_string):
     return int(size)  # 转换为整数
 
 
+def search_youtube(keywords):
+    YDL_OPTIONS = {
+        'format': 'bestaudio',
+        # 'noplaylist': 'True',
+        'proxy': 'socks5://127.0.0.1:1089',
+    }
+    with YoutubeDL(YDL_OPTIONS) as ydl:
+        video = ydl.extract_info(f"ytsearch:{keywords}", download=False)['entries'][0:5]
+        # video = ydl.extract_info(keywords, download=False)
+    if len(video) > 0:
+        ret = random.choice(video)
+        return ydl.sanitize_info(ret)
+    else:
+        return None
+
+
+def download_youtube(info, save_path):
+    url = info['original_url']
+    duration = info['duration']
+    if duration > 700:
+        start_second = 500
+        end_second = 530
+    else:
+        start_second = duration / 2 - 15
+        end_second = duration / 2 + 15
+        if start_second < 0:
+            start_second = 0
+        if end_second > duration:
+            end_second = duration
+    ydl_opts = {
+        'format': 'm4a/bestaudio/best',
+        'downloader': 'ffmpeg',
+        'download_ranges': download_range_func(None, [(start_second, end_second)]),
+        # ℹ️ See help(yt_dlp.postprocessor) for a list of available Postprocessors and their arguments
+        'postprocessors': [{  # Extract audio using ffmpeg
+            'key': 'FFmpegExtractAudio',
+            'preferredcodec': 'mp3',
+        }],
+        'outtmpl': save_path,
+        'proxy': 'http://127.0.0.1:8889',
+    }
+    with yt_dlp.YoutubeDL(ydl_opts) as ydl:
+        info = ydl.extract_info(url, download=True)
+        # ℹ️ ydl.sanitize_info makes the info json-serializable
+        ret_info = ydl.sanitize_info(info)
+        ret_info['save_path'] = save_path
+        return ret_info
+
+
+def get_youtube(keywords, save_path):
+    info = search_youtube(keywords)
+    if info is None:
+        return None
+    else:
+        return download_youtube(info, save_path)
+
+
 def get_albums(keywords, config):
     target_srcs = [
         'kugou', 'kuwo', 'qqmusic', 'qianqian', 'fivesing',
@@ -73,10 +135,9 @@ def get_albums(keywords, config):
     return valid_albums
 
 
-def get_random_spit(songinfo):
+def get_random_spit(songinfo, save_path):
     d = Downloader(songinfo)
     d.start()
-    save_path = os.path.join(songinfo["savedir"], f"{songinfo['savename']}.{songinfo['ext']}")
     song = AudioSegment.from_mp3(save_path)
     # pydub does things in milliseconds
     length = len(song)
@@ -91,6 +152,17 @@ def get_random_spit(songinfo):
     return save_path
 
 
+def download_random(keywords, config, save_path):
+    albums = get_albums(keywords, config)
+    if len(albums) == 0:
+        return None
+    album = random.choice(albums)
+    get_random_spit(album, save_path=save_path)
+
+
 if __name__ == '__main__':
-    config = {'logfilepath': 'musicdl.log', 'downloaded': 'downloaded', 'search_size_per_source': 5, 'proxies': {}}
-    get_albums('tayler', config)
+    # config = {'logfilepath': 'musicdl.log', 'downloaded': 'downloaded', 'search_size_per_source': 5, 'proxies': {}}
+    # infos = get_albums('李荣浩', config)
+    # print(infos)
+    info = search_youtube('李荣浩')
+    download_youtube(info, "downloaded")

requirements.txt

@@ -14,4 +14,5 @@ tqdm
 librosa
 pydub
 musicdl
-spleeter
+spleeter
+yt-dlp