Ver 1.4.3: deps, fish-audio-preprocess integrate

.gitignore

@@ -169,4 +169,5 @@ Info.plist
 /*.wav
 /*.flac
 output/
-*.exe
+*.exe
+fish_audio_preprocess

README.md

@@ -34,9 +34,7 @@
 conda create -n pyqt python=3.10
 conda activate pyqt
 pip install pdm
-# for windows nuitka build
-pip install nuitka
-pdm install
+pdm sync
 pdm run build.py
 ```
 
@@ -56,3 +54,17 @@ dist\fish.exe
 # linux
 dist/fish
 ```
+
+## Tips
+
+### Use FAP(Fish Audio Preprocess)
+
+1. To install the dependencies for the **FAP Toolkit**, run the following command in this project root:
+
+```bash
+pip install -r extra-requirements.txt
+```
+
+2. Open the GUI and ensure the Python path is correctly configured to the Python environment where you ran the above command.
+
+3. You are now ready to use the toolkit.

build.py

@@ -28,6 +28,7 @@
         "--output-filename=fish",
         "--include-data-dir=assets=assets",
         "--include-data-dir=locales=locales",
+        "--include-data-files=fish_audio_preprocess=fish_audio_preprocess/=**/*.py",
         "--windows-console-mode=disable",
         "--enable-plugins=pkg-resources",
         "--enable-plugins=pyqt6",

extra-requirements.txt

@@ -0,0 +1,12 @@
+tqdm>=4.64.1
+demucs>=4.0.0
+loguru>=0.6.0
+pyloudnorm>=0.1.1
+matplotlib>=3.6.2
+librosa>=0.9.0
+richuru>=0.1.1
+praat-parselmouth>=0.4.3
+click>=8.0.0
+faster-whisper
+funasr
+modelscope

fish/chat.py

@@ -830,7 +830,10 @@ async def send_message_async(self):
 
         if self.state.added_sysaudio is False and len(self.system_audios) > 0:
             self.state.added_sysaudio = True
-            sys_codes = [await agent.get_codes(audio) for audio in self.system_audios]
+            sys_codes = await asyncio.gather(
+                *[agent.get_codes(audio) for audio in self.system_audios]
+            )
+
             for sys_code in sys_codes:
                 self.state.append_to_chat_ctx(
                     ServeVQPart(codes=sys_code), role="system"
@@ -922,7 +925,9 @@ def run(self):
         is_voice_clickable=False,
         voice_duration=12,
     )
-    chat_widget.add_message("语音消息示例2", is_sender=False, is_voice=True, voice_duration=3)
+    chat_widget.add_message(
+        "语音消息示例2", is_sender=False, is_voice=True, voice_duration=3
+    )
 
     chat_widget.show()
 

fish/config.py

@@ -26,11 +26,13 @@ class Config:
     llm_url: str = "http://localhost:8080/v1/chat"
     proxy_url: str = "http://127.0.0.1:7890"
 
-    system_prompt: str = 'You are a voice assistant created by Fish Audio, offering end-to-end\
+    system_prompt: str = (
+        'You are a voice assistant created by Fish Audio, offering end-to-end\
  voice interaction for a seamless user experience. You are required to first transcribe\
  the user\'s speech, then answer it in the following format: "Question: [USER_SPEECH]\n\
   \nResponse: [YOUR_RESPONSE]\n"\u3002You are required to use the following voice\
  in this conversation.'
+    )
 
     ref_id: str = ""
     save_path: str = str(Path.cwd() / "output")

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "fish-speech-gui"
-version = "1.4.2"
+version = "1.4.3"
 description = "fish-speech comfortable GUI"
 readme = "README.md"
 requires-python = "<3.12,>=3.10"
@@ -17,14 +17,20 @@ classifiers = [
 
 dependencies = [
     "noisereduce>=2.0.1",
-    "sounddevice>=0.4.6",
+    "sounddevice==0.4.6",
+    "soundfile==0.12.1",
+    "pyaudio==0.2.14",
     "pyyaml>=6.0",
-    "PyQt6>=6.5.0",
+    "PyQt6==6.5.0",
+    "pyqt6-qt6==6.5.1",
+    "pyqt6-sip==13.5.1",
     "pyqtdarktheme==2.1.0",
     "requests>=2.31.0",
-    "librosa==0.9.2",
     "pydantic>=2.9.1",
     "ormsgpack>=1.5.0",
+    "numpy==1.24.3",
+    "psutil>=6.1.0",
+    "loguru>=0.7.2",
 ]
 
 [project.urls]
@@ -46,8 +52,9 @@ packages = { find = { include = ["fish*"], exclude = ["assets", "locales"] } }
 dev = [
     "isort>=5.12.0",
     "black>=23.3.0",
-    "pytest>=7.3.1",
+    "pytest==7.3.1",
     "pyinstaller>=5.11.0",
+    "nuitka>=2.4.11",
 ]
 
 [build-system]