blackteay/GPT-SoVITS
1
0
Fork 0
mirror of https://github.com/RVC-Boss/GPT-SoVITS.git synced 2025-04-06 03:57:44 +08:00
Code Issues Packages Projects Releases Wiki Activity

调整项目结构，修复随机采样bug

Browse Source
This commit is contained in:
Downupanddownup 2024-04-28 15:49:05 +08:00
parent 6cb3c15448
commit 27325f4cf9
3 changed files with 15 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

0
Ref_Audio_Selector/tool/model_manager.py → Ref_Audio_Selector/common/model_manager.py
View File

14
Ref_Audio_Selector/ref_audio_selector_webui.py
View File

@ -6,8 +6,8 @@ import gradio as gr
from Ref_Audio_Selector.config_param.log_config import logger from Ref_Audio_Selector.config_param.log_config import logger
import Ref_Audio_Selector.tool.model_manager as model_manager import Ref_Audio_Selector.common.model_manager as model_manager
import Ref_Audio_Selector.tool.audio_similarity as audio_similarity import Ref_Audio_Selector.tool.audio_sample as audio_sample
import Ref_Audio_Selector.tool.audio_inference as audio_inference import Ref_Audio_Selector.tool.audio_inference as audio_inference
import Ref_Audio_Selector.tool.audio_config as audio_config import Ref_Audio_Selector.tool.audio_config as audio_config
import Ref_Audio_Selector.tool.audio_check as audio_check import Ref_Audio_Selector.tool.audio_check as audio_check
@ -58,12 +58,12 @@ def convert_from_list(text_work_space_dir, text_role, text_list_input):
ref_audio_all = os.path.join(base_role_dir, ref_audio_all = os.path.join(base_role_dir,
params.list_to_convert_reference_audio_dir) params.list_to_convert_reference_audio_dir)
time_consuming, _ = time_util.time_monitor(audio_similarity.convert_from_list)(text_list_input, ref_audio_all) time_consuming, _ = time_util.time_monitor(audio_sample.convert_from_list)(text_list_input, ref_audio_all)
text_convert_from_list_info = f"耗时：{time_consuming:0.1f}秒；转换成功：生成目录{ref_audio_all}" text_convert_from_list_info = f"耗时：{time_consuming:0.1f}秒；转换成功：生成目录{ref_audio_all}"
text_sample_dir = ref_audio_all text_sample_dir = ref_audio_all
# audio_similarity.convert_from_list(text_list_input, ref_audio_all) # audio_sample.convert_from_list(text_list_input, ref_audio_all)
except Exception as e: except Exception as e:
logger.error("发生异常: \n%s", traceback.format_exc()) logger.error("发生异常: \n%s", traceback.format_exc())
text_convert_from_list_info = f"发生异常：{e}" text_convert_from_list_info = f"发生异常：{e}"
@ -92,11 +92,11 @@ def start_similarity_analysis(work_space_dir, sample_dir, base_voice_path, need_
p_similarity = Popen(cmd, shell=True) p_similarity = Popen(cmd, shell=True)
p_similarity.wait() p_similarity.wait()
similarity_list = audio_similarity.parse_similarity_file(similarity_file) similarity_list = audio_sample.parse_similarity_file(similarity_file)
if need_similarity_output: if need_similarity_output:
similarity_file_dir = os.path.join(similarity_dir, base_voice_file_name) similarity_file_dir = os.path.join(similarity_dir, base_voice_file_name)
audio_similarity.copy_and_move(similarity_file_dir, similarity_list) audio_sample.copy_and_move(similarity_file_dir, similarity_list)
p_similarity = None p_similarity = None
return similarity_list, similarity_file, similarity_file_dir return similarity_list, similarity_file, similarity_file_dir
@ -137,7 +137,7 @@ def sample(text_work_space_dir, text_role, text_sample_dir, text_base_voice_path
if similarity_list is None: if similarity_list is None:
raise Exception("相似度分析失败") raise Exception("相似度分析失败")
audio_similarity.sample(ref_audio_dir, similarity_list, slider_subsection_num, slider_sample_num) audio_sample.sample(ref_audio_dir, similarity_list, slider_subsection_num, slider_sample_num)
except Exception as e: except Exception as e:
logger.error("发生异常: \n%s", traceback.format_exc()) logger.error("发生异常: \n%s", traceback.format_exc())

13
Ref_Audio_Selector/tool/audio_similarity.py → Ref_Audio_Selector/tool/audio_sample.py
View File

@ -85,11 +85,14 @@ def sample(output_audio_dir, similarity_list, subsection_num, sample_num):
end = (i + 1) * step end = (i + 1) * step
end = min(end, len(similarity_list)) # 防止最后一段越界 end = min(end, len(similarity_list)) # 防止最后一段越界
num = min(sample_num, len(similarity_list[start:end])) # 创建子列表
subsection = similarity_list[start:end]
# 在子列表上随机打乱
random.shuffle(subsection)
# 随机采样 # 从打乱后的子列表中抽取相应数量的个体
random.shuffle(similarity_list[start:end]) num = min(sample_num, len(subsection))
sampled_subsection = similarity_list[start:start + num] sampled_subsection = subsection[:num]
# 创建并进入子目录 # 创建并进入子目录
subdir_name = f'emotion_{i + 1}' subdir_name = f'emotion_{i + 1}'
@ -143,7 +146,7 @@ def copy_and_move(output_audio_directory, similarity_scores):
for item in similarity_scores: for item in similarity_scores:
# 构造新的文件名 # 构造新的文件名
base_name = os.path.basename(item['wav_path'])[:-4] # 去掉.wav扩展名 base_name = os.path.basename(item['wav_path'])[:-4] # 去掉.wav扩展名
new_name = f"{item['score']*10000:04.0f}-{base_name}.wav" new_name = f"{item['score'] * 10000:04.0f}-{base_name}.wav"
# 新文件的完整路径 # 新文件的完整路径
new_path = os.path.join(output_audio_directory, new_name) new_path = os.path.join(output_audio_directory, new_name)