feat: dual-layer realtime transcription via RealtimeSTT

.github/workflows/release.yml

@@ -0,0 +1,50 @@
+name: Release
+
+on:
+  push:
+    tags:
+      - 'v*.*.*'
+
+permissions:
+  contents: write
+
+jobs:
+  build-and-release:
+    runs-on: windows-latest
+
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+
+      - name: Set up Python 3.11
+        uses: actions/setup-python@v5
+        with:
+          python-version: '3.11'
+          cache: pip
+
+      - name: Install dependencies
+        run: pip install -r requirements.txt pyinstaller
+
+      - name: Update installer version
+        shell: pwsh
+        run: |
+          $version = "${{ github.ref_name }}".TrimStart("v")
+          (Get-Content installer.iss) -replace 'AppVersion=.*', "AppVersion=$version" | Set-Content installer.iss
+
+      - name: Build with PyInstaller
+        shell: cmd
+        run: build.bat
+
+      - name: Install Inno Setup
+        shell: pwsh
+        run: choco install innosetup --yes --no-progress
+
+      - name: Build installer
+        shell: pwsh
+        run: '& "C:\Program Files (x86)\Inno Setup 6\ISCC.exe" installer.iss'
+
+      - name: Create GitHub Release
+        uses: softprops/action-gh-release@v2
+        with:
+          files: installer_output/HearsaySetup.exe
+          generate_release_notes: true

build.bat

@@ -14,9 +14,24 @@ pyinstaller --noconfirm --onedir --windowed ^
     --hidden-import "sounddevice" ^
     --hidden-import "customtkinter" ^
     --hidden-import "pystray" ^
+    --hidden-import "RealtimeSTT" ^
+    --hidden-import "silero_vad" ^
+    --hidden-import "webrtcvad" ^
+    --hidden-import "onnxruntime" ^
+    --hidden-import "scipy" ^
+    --hidden-import "soundfile" ^
+    --hidden-import "torch" ^
+    --hidden-import "torchaudio" ^
     --collect-all "customtkinter" ^
     --collect-all "faster_whisper" ^
     --collect-all "ctranslate2" ^
+    --collect-all "RealtimeSTT" ^
+    --collect-all "silero_vad" ^
+    --collect-all "onnxruntime" ^
+    --collect-all "scipy" ^
+    --collect-all "soundfile" ^
+    --collect-all "torch" ^
+    --collect-all "torchaudio" ^
     src\hearsay\__main__.py
 
 echo.

requirements.txt

@@ -1,7 +1,14 @@
 faster-whisper>=1.0.0
+RealtimeSTT>=1.0.0
+silero-vad>=5.1
 PyAudioWPatch>=0.2.12
 sounddevice>=0.4.6
 numpy>=1.24.0
 customtkinter>=5.2.0
 pystray>=0.19.5
 Pillow>=10.0.0
+nvidia-cublas-cu12>=12.0
+nvidia-cuda-runtime-cu12>=12.0
+transformers>=4.23.0
+torch>=2.0.0
+keyboard>=0.13.5

src/hearsay/__main__.py

@@ -1,13 +1,23 @@
 """Entry point for Hearsay: python -m hearsay"""
 
+import multiprocessing
 import sys
 
 
 def main() -> None:
+    # RealtimeSTT spawns a child process (spawn start method) for the main
+    # transcription model; freeze_support is required for frozen/PyInstaller builds.
+    multiprocessing.freeze_support()
+
     from hearsay.utils.logging_setup import setup_logging
 
     setup_logging()
 
+    # Must run before any ctranslate2 / faster-whisper import on Windows
+    from hearsay.utils.cuda_dlls import register_nvidia_dlls
+
+    register_nvidia_dlls()
+
     from hearsay.app import HearsayApp
 
     app = HearsayApp()