Ramnathan Muthumanickam Email & Phone Number

Chennai, Tn, In

Chennai, Tamil Nadu, India

Chennai, Tamil Nadu, India

Chennai, Tamil Nadu, India

Voice Synthesis (Associated with Interverse)• Created a Python script to generate phonetically balanced audio data for 300+ popular movie/cartoon/game characters.• Built a multispeaker text-to-speech (TTS) model based on Conditional Variational Autoencoder with Adversarial Learning (VITS architecture)• Achieved voice fusion from two unique speakers and created a new voice incorporating both speaker characteristics. Interactive Event Bot (Associated with Zoi Meet)• Trained a speech-to-text model (STT) using LSTM architecture to convert the user's speech input into text.• Created a Rasa NLU workflow by adding rules, fallbacks, and customized responses to make the bot more interactive and intelligent.Automatic Speech Recognition for Korean Language• Collecting requirements, processing data, and training it on Fairseq (Transformers based architecture)• Added a customized language model using n-grams to reduce the WER% of the model.• Dockerized the ASR model and then deployed on AWS EC2 instance.

India

Voicebot Product• Scripted a Micro-service architecture using Python-Flask.• Data Preparation Python-pandas to filter out outlier audio files/broken audio files.• Created a recurrent neural network and extracts audio signal features as input. • Augmentation of training data by adding real-world sound/noise to reduce the error rate of the model.Resume Extraction Automation• Scripted a python code to take resumes of any format as input source and extracting the text using Textract and Regular Expressions• Worked as an individual contributor to the development of filtering modules and Demo-ed features to the stakeholders periodically.Know Your Customer (KYC) Extraction• Made major improvements in the image preprocessing module by adding gamma correction to reduce image reflections of image using PIL module and testing different thresholding methods to achieve an accurate extraction in Tesseract.• Created an Image Tampering Detection module to determine whether the image is tampered or not, and also provides the name of the software used with the area where the image has been tampered with.Automatic Number Plate Recognition System (ANPR)• Detect and extract the number from the license plate, and identify vehicle type and model for a parking slot. • Data collected using web scrapping and labelled/annotated using labelme tool.• Built and trained the CNN using the pre-trained YOLO for object detection.• Designed model pipelines and hosted it as a web service using flask.Chatbot Product• Requirement gathering and text preprocessing using spaCy and BERT for word embedding.• Introduced AI to the chatbot application by integrating that with RNN (lstm) model to predict the next user query based on the chat flow.