set proxy 并没有在SocketServer.py脚本的参数中体现

.gitignore

@@ -0,0 +1,4 @@
+/.history/*
+/TTS/models/*
+/ASR/resources/models/*
+/SentimentEngine/models/*

GPT/GPTService-openai.py

@@ -0,0 +1,97 @@
+import logging
+import os
+import time
+
+import GPT.machine_id
+import GPT.tune as tune
+
+
+class GPTService():
+    def __init__(self, args):
+        logging.info('Initializing ChatGPT Service...')
+        self.chatVer = args.chatVer
+
+        self.tune = tune.get_tune(args.character, args.model)
+
+        self.counter = 0
+
+        self.brainwash = args.brainwash
+
+        if self.chatVer == 1:
+            from revChatGPT.V1 import Chatbot
+            config = {}
+            if args.accessToken:
+                logging.info('Try to login with access token.')
+                config['access_token'] = args.accessToken
+
+            else:
+                logging.info('Try to login with email and password.')
+                config['email'] = args.email
+                config['password'] = args.password
+            config['paid'] = args.paid
+            config['model'] = args.model
+            if type(args.proxy) == str:
+                config['proxy'] = args.proxy
+
+            self.chatbot = Chatbot(config=config)
+            logging.info('WEB Chatbot initialized.')
+
+
+        elif self.chatVer == 3:
+            mach_id = GPT.machine_id.get_machine_unique_identifier()
+            from revChatGPT.V3 import Chatbot
+            if args.APIKey:
+                logging.info('you have your own api key. Great.')
+                api_key = args.APIKey
+            else:
+                logging.info('using custom API proxy, with rate limit.')
+                os.environ['API_URL'] = "https://api.geekerwan.net/chatgpt2"
+                api_key = mach_id
+
+            self.chatbot = Chatbot(api_key=api_key, proxy=args.proxy, system_prompt=self.tune)
+            logging.info('API Chatbot initialized.')
+
+    def ask(self, text):
+        stime = time.time()
+        if self.chatVer == 3:
+            prev_text = self.chatbot.ask(text)
+
+        # V1
+        elif self.chatVer == 1:
+            for data in self.chatbot.ask(
+                    self.tune + '\n' + text
+            ):
+                prev_text = data["message"]
+
+        logging.info('ChatGPT Response: %s, time used %.2f' % (prev_text, time.time() - stime))
+        return prev_text
+
+    def ask_stream(self, text):
+        prev_text = ""
+        complete_text = ""
+        stime = time.time()
+        if self.counter % 5 == 0 and self.chatVer == 1:
+            if self.brainwash:
+                logging.info('Brainwash mode activated, reinforce the tune.')
+            else:
+                logging.info('Injecting tunes')
+            asktext = self.tune + '\n' + text
+        else:
+            asktext = text
+        self.counter += 1
+        for data in self.chatbot.ask(asktext) if self.chatVer == 1 else self.chatbot.ask_stream(text):
+            message = data["message"][len(prev_text):] if self.chatVer == 1 else data
+
+            if ("。" in message or "！" in message or "？" in message or "\n" in message) and len(complete_text) > 3:
+                complete_text += message
+                logging.info('ChatGPT Stream Response: %s, @Time %.2f' % (complete_text, time.time() - stime))
+                yield complete_text.strip()
+                complete_text = ""
+            else:
+                complete_text += message
+
+            prev_text = data["message"] if self.chatVer == 1 else data
+
+        if complete_text.strip():
+            logging.info('ChatGPT Stream Response: %s, @Time %.2f' % (complete_text, time.time() - stime))
+            yield complete_text.strip()

GPT/GPTService.py

@@ -1,20 +1,28 @@
 import logging
 import os
 import time
-
 import GPT.machine_id
 import GPT.tune as tune
-
+import requests
+import re
 
 class GPTService():
+    #替换一些听起来不太顺畅的文字
+    def _replace_str(self, desstr):
+        desstr = desstr.replace("！是派蒙！", "！我是派蒙！")
+        desstr = desstr.replace("Paimon", "派蒙")
+        desstr = desstr.replace("这是派蒙", "我是派蒙")
+        return desstr
+
+    #过滤特殊字符, 过滤除中英文, 数字，半角（, .,?:"!），全角（，。？：“！）以外的其他字符，用于防止颜文字
+    def _filter_str(self, desstr):
+        return ''.join(re.findall(u'[\u4e00-\u9fa5a-zA-Z0-9\u002E\u002C\u003B\u003A\u0022\u0021\u0020\u003F\u0027\uFF0C\u3002\uFF1F\uFF1A\u201C\uFF01]', desstr)) 
+
     def __init__(self, args):
         logging.info('Initializing ChatGPT Service...')
         self.chatVer = args.chatVer
-
         self.tune = tune.get_tune(args.character, args.model)
-
         self.counter = 0
-
         self.brainwash = args.brainwash
 
         if self.chatVer == 1:
@@ -23,7 +31,6 @@ def __init__(self, args):
             if args.accessToken:
                 logging.info('Try to login with access token.')
                 config['access_token'] = args.accessToken
-
             else:
                 logging.info('Try to login with email and password.')
                 config['email'] = args.email
@@ -36,34 +43,37 @@ def __init__(self, args):
             self.chatbot = Chatbot(config=config)
             logging.info('WEB Chatbot initialized.')
 
-
         elif self.chatVer == 3:
             mach_id = GPT.machine_id.get_machine_unique_identifier()
             from revChatGPT.V3 import Chatbot
             if args.APIKey:
                 logging.info('you have your own api key. Great.')
                 api_key = args.APIKey
             else:
-                logging.info('using custom API proxy, with rate limit.')
-                os.environ['API_URL'] = "https://api.geekerwan.net/chatgpt2"
+                logging.info('using custom API proxy, with rate limit.')                
+                #os.environ['API_URL'] = "http://localhost:8000/v1/chat/completions"
+                os.environ['API_URL'] = "http://75.63.212.152:41841/v1/"
                 api_key = mach_id
-
-            self.chatbot = Chatbot(api_key=api_key, proxy=args.proxy, system_prompt=self.tune)
+            self.chatbot = Chatbot(engine=args.model, api_key=api_key, proxy=args.proxy, system_prompt=self.tune, max_tokens=16000)
             logging.info('API Chatbot initialized.')
 
     def ask(self, text):
+        print("--->ask text:" + text)
         stime = time.time()
         if self.chatVer == 3:
             prev_text = self.chatbot.ask(text)
-
         # V1
         elif self.chatVer == 1:
             for data in self.chatbot.ask(
                     self.tune + '\n' + text
             ):
                 prev_text = data["message"]
-
-        logging.info('ChatGPT Response: %s, time used %.2f' % (prev_text, time.time() - stime))
+        #logging.info('ChatGPT Response: %s, time used %.2f' % (prev_text, time.time() - stime))
+        print("--->response:" + prev_text)
+        prev_text = self._filter_str(prev_text)
+        print("--->filterred:" + prev_text)
+        prev_text = self._replace_str(prev_text)
+        print("--->replaced:" + prev_text)
         return prev_text
 
     def ask_stream(self, text):
@@ -81,7 +91,6 @@ def ask_stream(self, text):
         self.counter += 1
         for data in self.chatbot.ask(asktext) if self.chatVer == 1 else self.chatbot.ask_stream(text):
             message = data["message"][len(prev_text):] if self.chatVer == 1 else data
-
             if ("。" in message or "！" in message or "？" in message or "\n" in message) and len(complete_text) > 3:
                 complete_text += message
                 logging.info('ChatGPT Stream Response: %s, @Time %.2f' % (complete_text, time.time() - stime))
@@ -94,4 +103,4 @@ def ask_stream(self, text):
 
         if complete_text.strip():
             logging.info('ChatGPT Stream Response: %s, @Time %.2f' % (complete_text, time.time() - stime))
-            yield complete_text.strip()
+            yield complete_text.strip()

GPT/machine_id.py

@@ -6,6 +6,7 @@
 import uuid
 
 def get_machine_unique_identifier():
+    mac_address = None
     if platform.system() == "Windows":
         # Use the Windows Management Instrumentation (WMI) interface
         import wmi

SocketServer.py

@@ -62,10 +62,10 @@ def __init__(self, args):
         self.addr = None
         self.conn = None
         logging.info('Initializing Server...')
-        self.host = socket.gethostbyname(socket.gethostname())
+        self.host = "0.0.0.0"# 所有地址上面监听
         self.port = 38438
         self.s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
-        self.s.setsockopt(socket.SOL_SOCKET, socket.SO_SNDBUF, 10240000)
+        self.s.setsockopt(socket.SOL_SOCKET, socket.SO_SNDBUF, 1024000)
         self.s.bind((self.host, self.port))
         self.tmp_recv_file = 'tmp/server_received.wav'
         self.tmp_proc_file = 'tmp/server_processed.wav'

TTS-flask-paimon.py

@@ -0,0 +1,76 @@
+from flask import Flask, request, send_file
+from pydub import AudioSegment
+import sys
+import time
+import os
+import torch
+import logging
+import soundfile
+
+sys.path.append('TTS/vits')
+import TTS.vits.commons as commons
+import TTS.vits.utils as utils
+
+from TTS.vits.models import SynthesizerTrn
+from TTS.vits.text.symbols import symbols
+from TTS.vits.text import text_to_sequence
+
+app = Flask(__name__)
+
+logging.getLogger().setLevel(logging.INFO)
+logging.basicConfig(level=logging.INFO)
+
+cfg_path = "TTS/models/paimon6k.json"
+model_path = "TTS/models/paimon6k_390k.pth"
+
+def get_text(text, hps):
+    text_norm = text_to_sequence(text, hps.data.text_cleaners)
+    if hps.data.add_blank:
+        text_norm = commons.intersperse(text_norm, 0)
+    text_norm = torch.LongTensor(text_norm)
+    return text_norm
+
+
+class TTService():
+    def __init__(self, cfg, model, char, speed):
+        logging.info('Initializing TTS Service for %s...' % char)
+        self.hps = utils.get_hparams_from_file(cfg)
+        self.speed = speed
+        self.net_g = SynthesizerTrn(
+            len(symbols),
+            self.hps.data.filter_length // 2 + 1,
+            self.hps.train.segment_size // self.hps.data.hop_length,
+            **self.hps.model).cuda()
+        _ = self.net_g.eval()
+        _ = utils.load_checkpoint(model, self.net_g, None)
+
+    def read(self, text):
+        text = text.replace('~', '！')
+        stn_tst = get_text(text, self.hps)
+        with torch.no_grad():
+            x_tst = stn_tst.cuda().unsqueeze(0)
+            x_tst_lengths = torch.LongTensor([stn_tst.size(0)]).cuda()
+            audio = self.net_g.infer(x_tst, x_tst_lengths, noise_scale=.667, noise_scale_w=0.2, length_scale=self.speed)[0][
+                0, 0].data.cpu().float().numpy()
+        return audio
+
+    def read_save(self, text, filename, sr):
+        stime = time.time()
+        au = self.read(text)
+        soundfile.write(filename, au, sr)
+        logging.info('VITS Synth Done, time used %.2f' % (time.time() - stime))
+
+# initialize TTService with appropriate arguments
+tts = TTService(cfg=cfg_path, model=model_path, char="char_var", speed=1.0)
+
+@app.route('/v1/audio/speech', methods=['POST'])
+def post_text_to_audio():
+    text = request.json['input']
+    audio = tts.read(text)
+    soundfile.write('audio.wav', audio, 44100)  # You may need to adjust the sample rate
+    sound = AudioSegment.from_wav("audio.wav")
+    sound.export("audio.mp3", format="mp3")
+    return send_file('audio.mp3', mimetype='audio/mp3')
+
+if __name__ == '__main__':
+    app.run(host="0.0.0.0", port=5000)

TTS/TTService-cuda.py

@@ -0,0 +1,59 @@
+import sys
+import time
+
+sys.path.append('TTS/vits')
+
+import soundfile
+import os
+os.environ["PYTORCH_JIT"] = "0"
+import torch
+
+import TTS.vits.commons as commons
+import TTS.vits.utils as utils
+
+from TTS.vits.models import SynthesizerTrn
+from TTS.vits.text.symbols import symbols
+from TTS.vits.text import text_to_sequence
+
+import logging
+logging.getLogger().setLevel(logging.INFO)
+logging.basicConfig(level=logging.INFO)
+
+
+def get_text(text, hps):
+    text_norm = text_to_sequence(text, hps.data.text_cleaners)
+    if hps.data.add_blank:
+        text_norm = commons.intersperse(text_norm, 0)
+    text_norm = torch.LongTensor(text_norm)
+    return text_norm
+
+
+class TTService():
+    def __init__(self, cfg, model, char, speed):
+        logging.info('Initializing TTS Service for %s...' % char)
+        self.hps = utils.get_hparams_from_file(cfg)
+        self.speed = speed
+        self.net_g = SynthesizerTrn(
+            len(symbols),
+            self.hps.data.filter_length // 2 + 1,
+            self.hps.train.segment_size // self.hps.data.hop_length,
+            **self.hps.model).cuda()
+        _ = self.net_g.eval()
+        _ = utils.load_checkpoint(model, self.net_g, None)
+
+    def read(self, text):
+        text = text.replace('~', '！')
+        stn_tst = get_text(text, self.hps)
+        with torch.no_grad():
+            x_tst = stn_tst.cuda().unsqueeze(0)
+            x_tst_lengths = torch.LongTensor([stn_tst.size(0)]).cuda()
+            audio = self.net_g.infer(x_tst, x_tst_lengths, noise_scale=.667, noise_scale_w=0.2, length_scale=self.speed)[0][
+                0, 0].data.cpu().float().numpy()
+        return audio
+
+    def read_save(self, text, filename, sr):
+        stime = time.time()
+        au = self.read(text)
+        soundfile.write(filename, au, sr)
+        logging.info('VITS Synth Done, time used %.2f' % (time.time() - stime))
+