type-a / speechnet

Automatic Speech Recognition

SpeechNET

End-to-End Automatic Speech Recognition (Taken down unntil research is completed)

Architecture

This model is based on the state of the art end-to-end speech recognotion system by google reseach. But rather just havig a single network approach, it has multiple pipeline and an end search algorithm to choose the appropriate phenomes.

Getting Started

These instructions will get you a copy of the project up and running on your local machine for development and testing purposes. See deployment for notes on how to deploy the project on a live system. This code is tested in Ubuntu 16.04 and Windows 10 with Python 3.6

git clone https://github.com/golu-golu/speechnet/
cd speechnet
python main.py

Prerequisites

What things you need to install the software and how to install them

pip install -r requirements.txt

Installing

A step by step series of examples that tell you have to get a development env running

To run with pretrained network

python main.py

To train locally and test

python train/train.py
python main.py

To test with single model file

python test_run.py -m  sound.mp3 

Running the tests

To run automated test run with

python main.py

Built With

• Tensorflow - Deep Leaning Framwork
• Scipy - Math :)
• Audio - Audio Processing