diff --git a/Ref_Audio_Selector/config/config_manager.py b/Ref_Audio_Selector/config/config_manager.py index 4c2e47c..6a7a188 100644 --- a/Ref_Audio_Selector/config/config_manager.py +++ b/Ref_Audio_Selector/config/config_manager.py @@ -44,4 +44,3 @@ def get_config(): if __name__ == '__main__': print(_config.print()) - print(_config.get_base('reference_audio_dir')) diff --git a/Ref_Audio_Selector/config/config_params.py b/Ref_Audio_Selector/config/config_params.py new file mode 100644 index 0000000..74c77c6 --- /dev/null +++ b/Ref_Audio_Selector/config/config_params.py @@ -0,0 +1,42 @@ +import Ref_Audio_Selector.config.config_manager as config_manager + +config = config_manager.get_config() + +# [Base] +# 参考音频目录 +reference_audio_dir = config.get_base('reference_audio_dir') + +# [AudioSample] +# list转换待选参考音频目录 +list_to_convert_reference_audio_dir = config.get_audio_sample('list_to_convert_reference_audio_dir') +# 音频相似度目录 +audio_similarity_dir = config.get_audio_sample('audio_similarity_dir') + +# [Inference] +# 默认测试文本位置 +default_test_text_path = config.get_inference('default_test_text_path') +# 推理音频目录 +inference_audio_dir = config.get_inference('inference_audio_dir') +# 推理音频文本聚合目录 +inference_audio_text_aggregation_dir = config.get_inference('inference_audio_text_aggregation_dir') +# 推理音频情绪聚合目录 +inference_audio_emotion_aggregation_dir = config.get_inference('inference_audio_emotion_aggregation_dir') + +# [ResultCheck] +# asr输出文件 +asr_filename = config.get_result_check('asr_filename') +# 文本相似度输出目录 +text_similarity_output_dir = config.get_result_check('text_similarity_output_dir') +# 文本情绪平均相似度报告文件名 +text_emotion_average_similarity_report_filename = config.get_result_check('text_emotion_average_similarity_report_filename') +# 文本相似度按情绪聚合明细文件名 +text_similarity_by_emotion_detail_filename = config.get_result_check('text_similarity_by_emotion_detail_filename') +# 文本相似度按文本聚合明细文件名 +text_similarity_by_text_detail_filename = config.get_result_check('text_similarity_by_text_detail_filename') + +# [AudioConfig] +# 默认模板文件位置 +default_template_path = config.get_audio_config('default_template_path') +# 参考音频配置文件名 +reference_audio_config_filename = config.get_audio_config('reference_audio_config_filename') + diff --git a/Ref_Audio_Selector/ref_audio_selector_webui.py b/Ref_Audio_Selector/ref_audio_selector_webui.py index 954361a..7eeba39 100644 --- a/Ref_Audio_Selector/ref_audio_selector_webui.py +++ b/Ref_Audio_Selector/ref_audio_selector_webui.py @@ -8,7 +8,7 @@ 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.delete_inference_with_ref as delete_inference_with_ref import Ref_Audio_Selector.common.common as common -import Ref_Audio_Selector.config.config_manager as config_manager +import Ref_Audio_Selector.config.config_params as params from tools.i18n.i18n import I18nAuto from config import python_exec, is_half from tools import my_utils @@ -16,7 +16,6 @@ from tools.asr.config import asr_dict from subprocess import Popen i18n = I18nAuto() -config = config_manager.get_config() p_similarity = None p_asr = None @@ -32,7 +31,7 @@ def check_base_info(text_work_space_dir): # 从list文件，提取参考音频 def convert_from_list(text_work_space_dir, text_list_input): ref_audio_all = os.path.join(text_work_space_dir, - config.get_audio_sample('list_to_convert_reference_audio_dir')) + params.list_to_convert_reference_audio_dir) text_convert_from_list_info = f"转换成功：生成目录{ref_audio_all}" text_sample_dir = ref_audio_all try: @@ -51,7 +50,7 @@ def start_similarity_analysis(work_space_dir, sample_dir, base_voice_path, need_ similarity_list = None similarity_file_dir = None - similarity_dir = os.path.join(work_space_dir, config.get_audio_sample('audio_similarity_dir')) + similarity_dir = os.path.join(work_space_dir, params.audio_similarity_dir) os.makedirs(similarity_dir, exist_ok=True) base_voice_file_name = common.get_filename_without_extension(base_voice_path) @@ -82,7 +81,7 @@ def start_similarity_analysis(work_space_dir, sample_dir, base_voice_path, need_ # 基于一个基准音频，从参考音频目录中进行分段抽样 def sample(text_work_space_dir, text_sample_dir, text_base_voice_path, text_subsection_num, text_sample_num, checkbox_similarity_output): - ref_audio_dir = os.path.join(text_work_space_dir, config.get_base('reference_audio_dir')) + ref_audio_dir = os.path.join(text_work_space_dir, params.reference_audio_dir) text_sample_info = f"抽样成功：生成目录{ref_audio_dir}" try: check_base_info(text_work_space_dir) @@ -117,9 +116,9 @@ def sample(text_work_space_dir, text_sample_dir, text_base_voice_path, def model_inference(text_work_space_dir, text_model_inference_voice_dir, text_url, text_text, text_ref_path, text_ref_text, text_emotion, text_test_content): - inference_dir = os.path.join(text_work_space_dir, config.get_inference('inference_audio_dir')) - text_asr_audio_dir = os.path.join(inference_dir, - config.get_inference('inference_audio_text_aggregation_dir')) + inference_dir = os.path.join(text_work_space_dir, params.inference_audio_dir) + text_asr_audio_dir = os.path.join(inference_dir, + params.inference_audio_text_aggregation_dir) text_model_inference_info = f"推理成功：生成目录{inference_dir}" try: check_base_info(text_work_space_dir) @@ -202,7 +201,7 @@ def open_asr(asr_inp_dir, asr_opt_dir, asr_model, asr_model_size, asr_lang): output_dir_abs = os.path.abspath(asr_opt_dir) output_file_name = os.path.basename(asr_inp_dir) # 构造输出文件路径 - output_file_path = os.path.join(output_dir_abs, f'{config.get_result_check("asr_filename")}.list') + output_file_path = os.path.join(output_dir_abs, f'{params.asr_filename}.list') return output_file_path else: @@ -212,7 +211,7 @@ def open_asr(asr_inp_dir, asr_opt_dir, asr_model, asr_model_size, asr_lang): # 对asr生成的文件，与原本的文本内容，进行相似度分析 def text_similarity_analysis(text_work_space_dir, text_text_similarity_analysis_path): - similarity_dir = os.path.join(text_work_space_dir, config.get_result_check('text_similarity_output_dir')) + similarity_dir = os.path.join(text_work_space_dir, params.text_similarity_output_dir) text_text_similarity_analysis_info = f"相似度分析成功：生成目录{similarity_dir}" try: check_base_info(text_work_space_dir) @@ -289,7 +288,7 @@ def sync_ref_audio(text_work_space_dir, text_sync_ref_audio_dir, # 根据模板和参考音频目录，生成参考音频配置内容 def create_config(text_work_space_dir, text_template, text_sync_ref_audio_dir2): - config_file = os.path.join(text_work_space_dir, f'{config.get_audio_config("reference_audio_config_filename")}.json') + config_file = os.path.join(text_work_space_dir, f'{params.reference_audio_config_filename}.json') text_create_config_info = f"配置生成成功：生成文件{config_file}" try: check_base_info(text_work_space_dir) @@ -359,7 +358,7 @@ with gr.Blocks() as app: text_emotion.input(whole_url, [text_url, text_text, text_ref_path, text_ref_text, text_emotion], [text_whole_url]) gr.Markdown(value=i18n("2.2：配置待推理文本，一句一行，不要太多，10条即可")) - default_test_content_path = config.get_inference('default_test_text_path') + default_test_content_path = params.default_test_text_path text_test_content = gr.Text(label=i18n("请输入待推理文本路径"), value=default_test_content_path) gr.Markdown(value=i18n("2.3：启动推理服务，如果还没启动的话")) gr.Markdown(value=i18n("2.4：开始批量推理，这个过程比较耗时，可以去干点别的")) @@ -427,7 +426,7 @@ with gr.Blocks() as app: text_sync_inference_audio_dir], [text_sync_ref_info]) with gr.Accordion("第四步：生成参考音频配置文本", open=False): gr.Markdown(value=i18n("4.1：编辑模板")) - default_template_path = config.get_audio_config('default_template_path') + default_template_path = params.default_template_path default_template_content = common.read_file(default_template_path) text_template_path = gr.Text(label=i18n("模板文件路径"), value=default_template_path, interactive=False) text_template = gr.Text(label=i18n("模板内容"), value=default_template_content, lines=10) diff --git a/Ref_Audio_Selector/tool/asr/funasr_asr_multi_level_dir.py b/Ref_Audio_Selector/tool/asr/funasr_asr_multi_level_dir.py index 24753b5..3fbb021 100644 --- a/Ref_Audio_Selector/tool/asr/funasr_asr_multi_level_dir.py +++ b/Ref_Audio_Selector/tool/asr/funasr_asr_multi_level_dir.py @@ -3,10 +3,9 @@ import argparse import os import traceback -import Ref_Audio_Selector.config.config_manager as config_manager +import Ref_Audio_Selector.config.config_params as params from tqdm import tqdm from funasr import AutoModel -config = config_manager.get_config() path_asr = 'tools/asr/models/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch' path_vad = 'tools/asr/models/speech_fsmn_vad_zh-cn-16k-common-pytorch' @@ -85,7 +84,7 @@ def execute_asr_multi_level_dir(input_folder, output_folder, model_size, languag os.makedirs(output_dir_abs, exist_ok=True) # 构造输出文件路径 - output_file_path = os.path.join(output_dir_abs, f'{config.get_result_check("asr_filename")}.list') + output_file_path = os.path.join(output_dir_abs, f'{params.asr_filename}.list') # 将输出写入文件 with open(output_file_path, "w", encoding="utf-8") as f: diff --git a/Ref_Audio_Selector/tool/audio_inference.py b/Ref_Audio_Selector/tool/audio_inference.py index 0df6b26..efa4746 100644 --- a/Ref_Audio_Selector/tool/audio_inference.py +++ b/Ref_Audio_Selector/tool/audio_inference.py @@ -1,11 +1,9 @@ import os import requests import itertools -import Ref_Audio_Selector.config.config_manager as config_manager +import Ref_Audio_Selector.config.config_params as params from urllib.parse import urlparse, parse_qs, urlencode, urlunparse, quote -config = config_manager.get_config() - class URLComposer: def __init__(self, base_url, emotion_param_name, text_param_name, ref_path_param_name, ref_text_param_name): @@ -80,9 +78,9 @@ def generate_audio_files(url_composer, text_list, emotion_list, output_dir_path) os.makedirs(output_dir, exist_ok=True) # Create subdirectories for text and emotion categories - text_subdir = os.path.join(output_dir, config.get_inference('inference_audio_text_aggregation_dir')) + text_subdir = os.path.join(output_dir, params.inference_audio_text_aggregation_dir) os.makedirs(text_subdir, exist_ok=True) - emotion_subdir = os.path.join(output_dir, config.get_inference('inference_audio_emotion_aggregation_dir')) + emotion_subdir = os.path.join(output_dir, params.inference_audio_emotion_aggregation_dir) os.makedirs(emotion_subdir, exist_ok=True) # 计算笛卡尔积 diff --git a/Ref_Audio_Selector/tool/delete_inference_with_ref.py b/Ref_Audio_Selector/tool/delete_inference_with_ref.py index 3a34c49..74ff9d9 100644 --- a/Ref_Audio_Selector/tool/delete_inference_with_ref.py +++ b/Ref_Audio_Selector/tool/delete_inference_with_ref.py @@ -1,9 +1,8 @@ import os import shutil import Ref_Audio_Selector.common.common as common -import Ref_Audio_Selector.config.config_manager as config_manager +import Ref_Audio_Selector.config.config_params as params -config = config_manager.get_config() def remove_matching_audio_files_in_text_dir(text_dir, emotions_list): count = 0 @@ -53,8 +52,8 @@ def delete_emotion_subdirectories(emotion_dir, emotions_list): def sync_ref_audio(ref_audio_dir, inference_audio_dir): ref_audio_manager = common.RefAudioListManager(ref_audio_dir) ref_list = ref_audio_manager.get_ref_audio_list() - text_dir = os.path.join(inference_audio_dir, config.get_inference('inference_audio_text_aggregation_dir')) - emotion_dir = os.path.join(inference_audio_dir, config.get_inference('inference_audio_emotion_aggregation_dir')) + text_dir = os.path.join(inference_audio_dir, params.inference_audio_text_aggregation_dir) + emotion_dir = os.path.join(inference_audio_dir, params.inference_audio_emotion_aggregation_dir) delete_text_wav_num = remove_matching_audio_files_in_text_dir(text_dir, ref_list) delete_emotion_dir_num = delete_emotion_subdirectories(emotion_dir, ref_list) return delete_text_wav_num, delete_emotion_dir_num diff --git a/Ref_Audio_Selector/tool/text_comparison/asr_text_process.py b/Ref_Audio_Selector/tool/text_comparison/asr_text_process.py index bc5791f..49b27ba 100644 --- a/Ref_Audio_Selector/tool/text_comparison/asr_text_process.py +++ b/Ref_Audio_Selector/tool/text_comparison/asr_text_process.py @@ -3,11 +3,9 @@ import argparse from collections import defaultdict from operator import itemgetter import Ref_Audio_Selector.tool.text_comparison.text_comparison as text_comparison -import Ref_Audio_Selector.config.config_manager as config_manager +import Ref_Audio_Selector.config.config_params as params import Ref_Audio_Selector.common.common as common -config = config_manager.get_config() - def parse_asr_file(file_path): output = [] @@ -100,19 +98,19 @@ def process(asr_file_path, output_dir, similarity_enlarge_boundary): average_similarity_list = calculate_average_similarity_by_emotion(records) average_similarity_file = os.path.join(output_dir, - f'{config.get_result_check("text_emotion_average_similarity_report_filename")}.txt') + f'{params.text_emotion_average_similarity_report_filename}.txt') average_similarity_content = \ '\n'.join([f"{item['average_similarity_score']}|{item['emotion']}" for item in average_similarity_list]) common.write_text_to_file(average_similarity_content, average_similarity_file) emotion_detail_list = group_and_sort_by_field(records, 'emotion') - emotion_detail_file = os.path.join(output_dir, f'{config.get_result_check("emotion_group_detail")}.txt') + emotion_detail_file = os.path.join(output_dir, f'{params.text_similarity_by_emotion_detail_filename}.txt') format_list_to_text(emotion_detail_list, emotion_detail_file) original_text_detail_list = group_and_sort_by_field(records, 'original_text') - original_text_detail_file = os.path.join(output_dir, f'{config.get_result_check("text_group_detail")}.txt') + original_text_detail_file = os.path.join(output_dir, f'{params.text_similarity_by_text_detail_filename}.txt') format_list_to_text(original_text_detail_list, original_text_detail_file) print('文本相似度分析完成。')