Merge pull request Ikaros-521#137 from Ikaros-521/owner

新增本地问答音频库，机制类似点歌，匹配命中后直接播放；新增自定义上下限音频随机变速(GUI未适配)

README.md

@@ -76,6 +76,7 @@ AI Vtuber是一个由 ChatterBot/GPT/Claude/langchain本地or云端/chatglm/text
     - [3.ERROR: Conmunicate.`__init__`() got an unexpected keyword argument 'text'](#3error-conmunicate__init__-got-an-unexpected-keyword-argument-text)
     - [4.File ".\flask_api_full_song.py", line 38, in wav2wav  out_audio, out_sr = svc_model.infer(spk, tran, raw_path) ValueError: too many values to unpack (expected 2)](#4file-%5Cflask_api_full_songpy-line-38-in-wav2wav--out_audio-out_sr--svc_modelinferspk-tran-raw_path-valueerror-too-many-values-to-unpack-expected-2)
 - [♏️补充](#%E8%A1%A5%E5%85%85)
+  - [b站直播监听](#b%E7%AB%99%E7%9B%B4%E6%92%AD%E7%9B%91%E5%90%AC)
   - [<span id="dy">抖音弹幕获取</span>](#span-iddy%E6%8A%96%E9%9F%B3%E5%BC%B9%E5%B9%95%E8%8E%B7%E5%8F%96span)
   - [快手弹幕获取](#%E5%BF%AB%E6%89%8B%E5%BC%B9%E5%B9%95%E8%8E%B7%E5%8F%96)
   - [ChatGPT](#chatgpt)
@@ -89,6 +90,7 @@ AI Vtuber是一个由 ChatterBot/GPT/Claude/langchain本地or云端/chatglm/text
   - [Live2D](#live2d)
   - [Stable Diffusion](#stable-diffusion)
   - [VITS-fast-fine-tuning](#vits-fast-fine-tuning)
+  - [DDSP-SVC](#ddsp-svc)
   - [so-vits-svc](#so-vits-svc)
   - [聊天模式 百度](#%E8%81%8A%E5%A4%A9%E6%A8%A1%E5%BC%8F-%E7%99%BE%E5%BA%A6)
 - [🀅开发&项目相关](#%F0%9F%80%85%E5%BC%80%E5%8F%91%E9%A1%B9%E7%9B%AE%E7%9B%B8%E5%85%B3)
@@ -105,6 +107,8 @@ AI Vtuber是一个由 ChatterBot/GPT/Claude/langchain本地or云端/chatglm/text
 - [⭐️Star 经历](#%EF%B8%8Fstar-%E7%BB%8F%E5%8E%86)
 - [🤝贡献](#%E8%B4%A1%E7%8C%AE)
   - [🎉鸣谢](#%E9%B8%A3%E8%B0%A2)
+  - [💸投资方](#%E6%8A%95%E8%B5%84%E6%96%B9)
+  - [🙌赞助](#%E8%B5%9E%E5%8A%A9)
 
 <!-- END doctoc generated TOC please keep comment here to allow auto update -->
 
@@ -1091,6 +1095,8 @@ cmd输入命令即可：`doctoc /path/to/file`
 ### 2023-07-19
 - GUI适配DDSP-SVC；补充说明文档
 - 优化audio中vits接口的设计&优化代码减少冗余
+- 新增本地问答音频库，机制类似点歌，通过编辑音频文件名做为触发匹配关键词，优先级第二高，匹配命中后直接会进行匹配音频的播放。
+- 新增自定义上下限音频随机变速，对普通音频播放和文案音频播放可以进行随机变速配置。(GUI未适配)
 
 </details>
 

config.json

@@ -6,7 +6,6 @@
   "before_prompt": "请简要回复:",
   "after_prompt": "",
   "commit_log_type": "回答",
-  "local_qa": false,
   "filter": {
     "before_must_str": [],
     "after_must_str": [],
@@ -57,6 +56,18 @@
     "local_vector_embedding_model": "sebastian-hofstaetter/distilbert-dot-tas_b-b256-msmarco"
   },
   "audio_synthesis_type": "edge-tts",
+  "audio_random_speed": {
+    "normal": {
+      "enable": false,
+      "speed_min": 0.8,
+      "speed_max": 1.5
+    },
+    "copywriting": {
+      "enable": false,
+      "speed_min": 0.8,
+      "speed_max": 1.5
+    }
+  },
   "vits": {
     "config_path": "D:\\GitHub_pro\\VITS-fast-fine-tuning\\inference\\finetune_speaker.json",
     "api_ip_port": "http://127.0.0.1:7860",
@@ -118,6 +129,16 @@
     "instruction_template": "Vicuna-v1.1",
     "your_name": "你"
   },
+  "local_qa": {
+    "text": {
+      "enable": true,
+      "file_path": "data/本地问答库.txt"
+    },
+    "audio": {
+      "enable": true,
+      "file_path": "out/本地问答音频/"
+    }
+  },
   "choose_song": {
     "enable": true,
     "start_cmd": "点歌 ",
@@ -150,6 +171,17 @@
   "copywriting": {
     "file_path": "data/copywriting/",
     "audio_path": "out/copywriting/",
+    "play_list": [
+      "伊卡日语介绍.wav",
+      "吐槽.wav"
+    ],
+    "audio_interval": 5.0,
+    "switching_interval": 5.0,
+    "random_play": true
+  },
+  "copywriting2": {
+    "file_path": "data/copywriting2/",
+    "audio_path": "out/copywriting2/",
     "play_list": [],
     "audio_interval": 5.0,
     "switching_interval": 5.0,

config.json.bak

@@ -6,7 +6,6 @@
   "before_prompt": "请简要回复:",
   "after_prompt": "",
   "commit_log_type": "回答",
-  "local_qa": false,
   "filter": {
     "before_must_str": [],
     "after_must_str": [],
@@ -57,11 +56,23 @@
     "local_vector_embedding_model": "sebastian-hofstaetter/distilbert-dot-tas_b-b256-msmarco"
   },
   "audio_synthesis_type": "edge-tts",
+  "audio_random_speed": {
+    "normal": {
+      "enable": false,
+      "speed_min": 0.8,
+      "speed_max": 1.5
+    },
+    "copywriting": {
+      "enable": false,
+      "speed_min": 0.8,
+      "speed_max": 1.5
+    }
+  },
   "vits": {
     "config_path": "D:\\GitHub_pro\\VITS-fast-fine-tuning\\inference\\finetune_speaker.json",
     "api_ip_port": "http://127.0.0.1:7860",
     "character": "ikaros",
-    "speed": 1
+    "speed": 1.0
   },
   "ddsp_svc": {
     "enable": false,
@@ -118,6 +129,16 @@
     "instruction_template": "Vicuna-v1.1",
     "your_name": "你"
   },
+  "local_qa": {
+    "text": {
+      "enable": true,
+      "file_path": "data/本地问答库.txt"
+    },
+    "audio": {
+      "enable": true,
+      "file_path": "out/本地问答音频/"
+    }
+  },
   "choose_song": {
     "enable": true,
     "start_cmd": "点歌 ",
@@ -136,7 +157,7 @@
     "ip": "127.0.0.1",
     "port": 7860,
     "negative_prompt": "ufsw, longbody, lowres, bad anatomy, bad hands, missing fingers, pubic hair,extra digit, fewer digits, cropped, worst quality, low quality",
-    "seed": -1,
+    "seed": -1.0,
     "styles": [],
     "cfg_scale": 7,
     "steps": 30,
@@ -150,9 +171,20 @@
   "copywriting": {
     "file_path": "data/copywriting/",
     "audio_path": "out/copywriting/",
+    "play_list": [
+      "伊卡日语介绍.wav",
+      "吐槽.wav"
+    ],
+    "audio_interval": 5.0,
+    "switching_interval": 5.0,
+    "random_play": true
+  },
+  "copywriting2": {
+    "file_path": "data/copywriting2/",
+    "audio_path": "out/copywriting2/",
     "play_list": [],
-    "audio_interval": 5,
-    "switching_interval": 5,
+    "audio_interval": 5.0,
+    "switching_interval": 5.0,
     "random_play": true
   },
   "header": {

data/copywriting2/测试文案.txt

@@ -0,0 +1,2 @@
+你好，我是文案2的测试文案。
+这是新机制，您可以配置2套文案，使其在文案1和文案2之间切换。

main.py

@@ -191,6 +191,9 @@ def init_config(self):
             # 日志
             self.captions_config = config.get("captions")
 
+            # 本地问答
+            self.local_qa_config = config.get("local_qa")
+
             # 过滤配置
             self.filter_config = config.get("filter")
 
@@ -258,6 +261,12 @@ def init_config(self):
             self.ui.label_captions_enable.setToolTip("是否启用字幕日志记录，字幕输出内容为当前合成播放的音频的文本")
             self.ui.label_captions_file_path.setToolTip("字幕日志存储路径")
 
+            # 本地问答
+            self.ui.label_local_qa_text_enable.setToolTip("是否启用本地问答文本匹配，完全命中设定的问题后，自动合成对应的回答")
+            self.ui.label_local_qa_text_file_path.setToolTip("本地问答文本数据存储路径")
+            self.ui.label_local_qa_audio_enable.setToolTip("是否启用本地问答音频匹配，部分命中音频文件名后，直接播放对应的音频文件")
+            self.ui.label_local_qa_audio_file_path.setToolTip("本地问答音频文件存储路径")
+
             self.ui.label_filter_before_must_str.setToolTip("弹幕过滤，必须携带的触发前缀字符串（任一）\n例如：配置#，那么就需要发送：#你好")
             self.ui.label_filter_after_must_str.setToolTip("弹幕过滤，必须携带的触发后缀字符串（任一）\n例如：配置。那么就需要发送：你好。")
             self.ui.label_filter_badwords_path.setToolTip("本地违禁词数据路径（你如果不需要，可以清空文件内容）")
@@ -480,6 +489,14 @@ def init_config(self):
                 self.ui.checkBox_captions_enable.setChecked(True)
             self.ui.lineEdit_captions_file_path.setText(self.captions_config['file_path'])
 
+            # 本地问答
+            if self.local_qa_config['text']['enable']:
+                self.ui.checkBox_local_qa_text_enable.setChecked(True)
+            self.ui.lineEdit_local_qa_text_file_path.setText(self.local_qa_config['text']['file_path'])
+            if self.local_qa_config['audio']['enable']:
+                self.ui.checkBox_local_qa_audio_enable.setChecked(True)
+            self.ui.lineEdit_local_qa_audio_file_path.setText(self.local_qa_config['audio']['file_path'])
+
             tmp_str = ""
             for tmp in self.filter_config['before_must_str']:
                 tmp_str = tmp_str + tmp + "\n"
@@ -953,6 +970,12 @@ def save(self):
             config_data["captions"]["enable"] = self.ui.checkBox_captions_enable.isChecked()
             config_data["captions"]["file_path"] = self.ui.lineEdit_captions_file_path.text()
 
+            # 本地问答
+            config_data["local_qa"]["text"]["enable"] = self.ui.checkBox_local_qa_text_enable.isChecked()
+            config_data["local_qa"]["text"]["file_path"] = self.ui.lineEdit_local_qa_text_file_path.text()
+            config_data["local_qa"]["audio"]["enable"] = self.ui.checkBox_local_qa_audio_enable.isChecked()
+            config_data["local_qa"]["audio"]["file_path"] = self.ui.lineEdit_local_qa_audio_file_path.text()
+
             # 通用多行分隔符
             separators = [" ", "\n"]
 
@@ -1392,6 +1415,7 @@ async def copywriting_synthetic_audio_main(self):
             # 刷新文案和音频列表
             self.copywriting_refresh_list()
 
+
     # 合成音频
     def copywriting_synthetic_audio(self):
         if self.running_flag != 1: