def get_class_other(tag_data):
"""
音乐分类
:param tag_data:
:return:
"""
print("其他-音乐分类中....")
result = [0, 0, 0, 0, 0, 0, 0, 0]
data = tag_data.iloc[:, 2:7]
estimator = joblib.load("music_clf.pkl")
for row in tqdm(data.itertuples()):
song_id = str(getattr(row, 'song_id'))
start_time = getattr(row, 'start_time')
end_time = getattr(row, 'end_time')
times = getattr(row, 'times')
tag = getattr(row,'tag')
if tag == 3:
times = end_time/times
elif tag == 4:
times = times * 0.7
# 判断是否已存在于wav文件夹中
wav_file = "E:\\musiclib\\wav\\" + str(song_id) + ".wav"
mp3_file = "E:\\musiclib\\mp3\\" + str(song_id) + ".mp3"
predict_file = "E:\\musiclib\\predict\\" + str(song_id) + ".txt"
mfcc_file = "E:\\musiclib\\mfcc\\" + str(song_id) + ".txt"
if music_classify.judge_exist(predict_file) == True:
y_predict = music_classify.Read_list_y(predict_file)
result[y_predict[0]] += times
continue
if music_classify.judge_exist(mfcc_file) == True:
x_predict = music_classify.Read_list_x(mfcc_file)
y_predict = estimator.predict(x_predict)
music_classify.Save_feature(y_predict,predict_file)
result[y_predict[0]] += times
continue
if music_classify.judge_exist(wav_file) == False:
if music_classify.judge_exist(mp3_file) == False:
# 否-尝试进行下载
if crawl_song.download_song(mp3_file, song_id) == False:
continue
if music_classify.mp3_to_wav(mp3_file, wav_file) == False:
continue
else:
if music_classify.mp3_to_wav(mp3_file, wav_file) == False:
continue
x_predict = [music_classify.get_mfcc_feature(wav_file)]
y_predict = estimator.predict(x_predict)
music_classify.Save_list(x_predict,mfcc_file)
music_classify.Save_list(y_predict,predict_file)
result[y_predict[0]] += times
return result
def save_predict(song_id):
save_file = "E:\\musiclib\\predict\\" + str(song_id) + ".txt"
if music_classify.judge_exist(save_file):
print("已存在，不必进行特征转换")
return
y_predict = []
y_predict.append(get_musicclass(song_id))
music_classify.Save_list(y_predict, save_file)
def save_mfcc(song_id):
target_wav = "E:\\musiclib\\wav\\"+str(song_id)+".wav"
save_file = "E:\\musiclib\\mfcc\\"+str(song_id)+".txt"
if music_classify.judge_exist(save_file):
print("已存在，不必进行特征转换")
return
x_predict = []
x_predict.append(music_classify.get_mfcc_feature(target_wav))
music_classify.Save_list(x_predict,save_file)
def get_tag2_class(tag2_data):
"""
音乐片段分类
:param tag2_data:
:return:
"""
print("片段-音乐分类中....")
result = [0,0,0,0,0,0,0,0]
data = tag2_data.iloc[:,2:7]
estimator = joblib.load("music_clf.pkl")
for row in tqdm(data.itertuples()):
song_id = str(getattr(row, 'song_id'))
start_time = getattr(row, 'start_time')
end_time = getattr(row, 'end_time')
times = getattr(row, 'times')
# 判断是否已存在于wav文件夹中
wav_file = "E:\\musiclib\\wav\\" + str(song_id) + ".wav"
mp3_file = "E:\\musiclib\\mp3\\" + str(song_id) + ".mp3"
if music_classify.judge_exist(wav_file) == False:
if music_classify.judge_exist(mp3_file) == False:
# 否-尝试进行下载
if crawl_song.download_song(mp3_file, song_id) == False:
continue
if music_classify.mp3_to_wav(mp3_file, wav_file) == False:
continue
else:
if music_classify.mp3_to_wav(mp3_file, wav_file) == False:
continue
edit_music.clip_music(wav_file,"clip.wav",start_time,end_time)
x_predict = [music_classify.get_mfcc_feature("clip.wav")]
y_predict = estimator.predict(x_predict)
result[y_predict[0]]+=times
return result
def work_tag1_lrc(user_id,tag1_data):
# 提取song_id
song_ids = tag1_data.iloc[:,2:7]
# print(song_ids)
# 遍历
for row in song_ids.itertuples():
song_id = str(getattr(row, 'song_id'))
times = getattr(row,'times')
# 判断是否已存在于lrc文件夹中
lrc_file = "E:\\musiclib\\lrc\\"+str(song_id)+".txt"
if music_classify.judge_exist(lrc_file)==False:
# 否-尝试进行下载
crawl_song.download_lrc(lrc_file,song_id)
# 是-拼接到用户词云txt
user_lrc_file = "user_data/cloud_"+str(user_id)+".txt"
for i in range(times):
edit_music.write_to_file(lrc_file,0,10000,user_lrc_file);