Mubarekethio/Voice-Recognition-Qafaraf-and-Amharic

Qafar-af and Amharic voice Command Recognition project to control the movement of wheelchair # Qafar-af Speech Command Recognition: Recognizing keywords This repository is demonstrates audio preprocessing and building a machine learning model for Qafaraf and Amharic speech command recognition. The audio files in the WAV format is organized in seven different folders for each command and each WAV audio file are 1 second long with 16khz sampling rate the following five voice command. - "Qembis", (start) - "Derreh", (back) - "Gurra", (turn left) - "Migda", (turn right) - "Solis", (stop) - "Noise" - "unknown" ## Qafar-af and Amharic Speech Command Recognition (QSCR) In this notebook the speech command recognition model is build and trained using the collected Qafar-af audio dataset. The Qafar-af Audio Dataset is contain seven folders that have five keyword, and noise an unknown class organized in seven class the using Qafar-af audio dataset a Multi-class audio classification involves categorizing audio samples into multiple predefined classes or categories. This can be accomplished using various machine learning and signal processing techniques. Here's a simplified step-by-step process: 1. Data Collection and Preprocessing Data Collection: Gather a diverse dataset containing audio samples for each class you want to classify. Preprocessing: Convert audio files into a standardized format (e.g., WAV, MP3) and extract features that represent the audio content effectively. Common features include Mel-Frequency Cepstral Coefficients (MFCCs), spectrograms, or other time-frequency representations. 2. Feature Extraction Extract Features: Utilize signal processing methods (like Fast Fourier Transform, Short-Time Fourier Transform, or other spectral analysis techniques) to extract numerical representations (features) from the audio files. These features should capture relevant information about the sound, such as its frequency content, amplitude, and temporal characteristics. 3. Model Selection and Training Model Selection: Choose an appropr …