[+] Windows setup guide

README.md

@@ -23,9 +23,33 @@ pip3 install -r requirements-mac.txt
 plaidml-setup
 ```
 
-4. Configure environment variables:
+4. Configure environment variables in the run script:
 
 ```sh
 export KERAS_BACKEND="plaidml.keras.backend"
 export tg_token="Your telegram token here"
 ```
+
+### Windows (CUDA)
+
+1. Setup Python
+
+```powershell
+python3.9 -m venv venv
+.\venv\Scripts\activate
+pip install -r requirements-win-cuda.txt
+```
+
+2. Install CUDA
+
+* Install NVIDIA Drivers: https://www.nvidia.com/drivers
+* Install CUDA **11.2** (for TensorFlow 2.7.0): https://developer.nvidia.com/cuda-toolkit-archive
+* Download cuDNN **8.1**: https://developer.nvidia.com/rdp/cudnn-archive
+  * Copy folders in `cudnn-11.2-windows-x64-v8.1.1.33\cuda` to `C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.2`
+* Restart IntelliJ IDEA
+
+3. Check Device List
+
+```shell
+python -c "from tensorflow.python.client import device_lib; print(device_lib.list_local_devices())"
+```

requirements-win-cuda.txt

@@ -0,0 +1,5 @@
+
+tensorflow==2.7.0
+inaSpeechSegmenter==0.6.8
+
+python-telegram-bot

src/bot.py

@@ -1,17 +1,12 @@
-import os
+import warnings
 from datetime import datetime
 from pathlib import Path
+
 from telegram import Update, Message
 from telegram.ext import Updater, CallbackContext, Dispatcher, CommandHandler, MessageHandler, \
     Filters
 
-os.environ['KERAS_BACKEND'] = "plaidml.keras.backend"
-
-import keras
-
-from inaSpeechSegmenter import *
 from ina_main import *
-import warnings
 
 warnings.filterwarnings("ignore")
 

src/ina_main.py

@@ -2,31 +2,19 @@ from __future__ import annotations
 
 import io
 import os
-import shutil
-import sys
 import tempfile
 import time
-import wave
+import warnings
-from dataclasses import dataclass
-
-from PIL import Image
 from subprocess import Popen, PIPE
 from typing import NamedTuple, Callable
 
 import matplotlib.pyplot as plt
 import numpy as np
 import scipy.io.wavfile
-from matplotlib.figure import Figure, Axes
+from PIL import Image
-from numpy import ndarray
-
-os.environ['KERAS_BACKEND'] = "plaidml.keras.backend"
-
-import keras
-from keras import backend
-
-import tensorflow as tf
 from inaSpeechSegmenter import *
 from inaSpeechSegmenter.segmenter import featGenerator
+from matplotlib.figure import Figure, Axes
 
 
 class ResultFrame(NamedTuple):
@@ -175,21 +163,30 @@ def get_result_percentages(result: Result) -> tuple[float, float, float]:
     return f, m, 1 - f - m
 
 
-# def test():
+def test():
-#     results: BatchResults = BatchResults(
+    # results: BatchResults = BatchResults(
-#         [Result([ResultFrame('female', 0.0, 10.48), ResultFrame('male', 10.48, 12.780000000000001)],
+    #     [Result([ResultFrame('female', 0.0, 10.48), ResultFrame('male', 10.48, 12.780000000000001)],
-#                 '../test.csv')],
+    #             '../test.csv')],
-#         1.7032792568206787, 1.7032792568206787, 1,
+    #     1.7032792568206787, 1.7032792568206787, 1,
-#         [('../test.csv', 0)])
+    #     [('../test.csv', 0)])
-#
+
-#     with draw_result('../test.mp3', results.results[0]) as buf:
+    warnings.filterwarnings("ignore")
-#         show_image_buffer(buf)
+    seg = Segmenter()
-#         print(get_result_percentages(results.results[0]))
+
-#
+    # Warmup run
-#     # seg = Segmenter()
+    results = process(seg, ['../test.mp3'])
-#     # print(process(seg, ['../test.mp3']))
+    print(results)
+
+    # Actual run
+    results = process(seg, ['../test.mp3'])
+    print(results)
+
+    # Draw results
+    with draw_result('../test.mp3', results.results[0]) as buf:
+        show_image_buffer(buf)
+        print(get_result_percentages(results.results[0]))
+
 
 if __name__ == '__main__':
-    # to_wav('../audio_tmp/2021-12-22 05-32 leph1art5.mp3', print)
+    test()
-    # test()
     pass