fbpx

BUILD A SPEECH RECOGNITION SYSTEM IN 5 WEEKS

2 Hours Online Session

Mondays and Wednesdays

Timings – 3 pm to 5 pm

Instructor: Dr. Ali Tahir

Course Duration: 5 weeks

Registration Fee: Rs. 20,000/-

Pre-requisites and System Requirements:

  • Basics of AI/ML (UG level)
  • Basics of NN
  • Programming experience in any language (C++, Python)
  • Experience with Linux shell commands (Bash scripting)
  • Ubuntu 14.04 and higher
  • 4GB RAM, Core i3 processor

COURSE OVERVIEW

In just five weeks, this online course from AI Lounge will walk you through the domain of speech recognition in detail. Do you want to know how Siri works or do you want to make Alexa of your own. Come join us in this speech recognition boot camp. Learn all the latest tools used in speech recognition including Acoustic Models, Urdu ASR, LSTM and many others. In AI-Lounge fashion, we combine lectures and hands on experience to optimize your learning. 

WEEK 1

Wednesday (Lecture)

Intro to Machine Learning

Friday (Hands-On)

– Intro to basic tools for ASR: Kaldi, etc.
– Linux shell and bash scripting
– Installation and compilation of Kaldi
-Video (screen recording)
– PDF document outlining step-by-step guide

WEEK 2

Monday (Lecture)

– Intro to Speech Recognition – Basics to Advanced
– Scope, Applications and Challenges
– Acoustic models, Bayes’ rule and Hidden Markov model

Wednesday (Hands-On)

– Basic English ASR model
– Dataset introduction and understanding (Kaldi Data Format)
– Downloading and preprocessing of data set

WEEK 3

Monday (Lecture)

– Language Model and Lexicon
– Phonemes and sub-word units and English pronunciation lexicon
– Understanding mathematics of language model

Wednesday (Hands-On)

– How to create Urdu language model and lexicon – Hands-on
– Data set collection and preprocessing
– Using SRILM toolkit for language model creation
– Perplexity optimization

WEEK 4

Monday (Lecture)

– Intro to Deep Neural Networks and mathematical definitions
– DNN-HMM hybrid architecture
– Feed forward neural networks and LSTM (for speech recognition)

Wednesday (Hands-On)

– ASR training using DNNs – hands-on
– Understanding workflow of DNN training phases in Kaldi
– Execution of different DNN training steps, their input,  output and    parameters.

WEEK 5

Monday (Lecture)

– Project Description
– Scope of projects
– Discussion about possible projects and issues/workload involved

Wednesday (Hands-On)

– Mid-project discussion
– Error messages/debugging and mathematical clarifications as required
– Project Submission on github repository

WEEK 6

Monday (Demo Day) – Extended Day

 

– Using Zoom, giving a demo of implemented project
– Question Answers Session
Confirm your Registration

Kindly confirm your registration by sending the payment receipt to “info@ai-lounge.com”

Account Title: DCUBE TECHNOLOGY PRIVATE LIMITED
IBAN: PK40MEZN0008020102445747
BIC: MEZNPKKA
Bank Name: Meezan Bank Limited, Bahria Heights