Speech Recognition

A project with deep learning networks that recognizes who owns a voice using libraries like scipy, sklearn, and python-speech-features

Setup This Project

Install Python2.7

• I recommend that you install Anaconda and install python 2.7 from the environments part of anaconda.
• After this installation, you can run the project by going to the project directory in your terminal and typing conda activate python2x

Install Libraries

• Switch to Python2x environment by typing conda activate Python2x
• Then you can install the modules by typing pip install -r requirements.txt in the terminal.
• Since you are running the project around the Python2x environment, you must also enable Python2x for use pip command

How this is working?

A program that recognizes the sound of the specified file using models using methods such as mfcc gmm. Code comments were entered as # comments on each line.

• train.py : Used for model audio files in trainData folder
• recognize.py : It is used to define a selected file or all files in the data folder.
• requirements.txt : Text file containing the necessary libraries
• ./models : The folder where train.py outputs and recognize.py uses. Here are the models of the trained audio files.
• ./Data : The files here use recognize.py. Here you should drop the file you want to define .
• ./trainingData : Sound files to be used for modeling are placed here.

For each audio file, it should be recorded as VoiceName-Integer/VoiceFile.wav Example: Melissa-005/Melissa.wav | Since the folder name is split, it should be saved like this.

You can get information by the link below. You might need to translate the page