Muhammad Danyal Khan

Call Centers have huge amount of audio data which can be used for achieving valuable business insights. This requires transcription of phone calls which is manually tedious task to perform. An effective Automated Speech Recognition system can accurately transcribe these calls, making it easier to search through call history for specific context and content with very less time and manual effort. With text transcription available calls can be automatically monitored, improving QoS through keyword… Show more

Call Centers have huge amount of audio data which can be used for achieving valuable business insights. This requires transcription of phone calls which is manually tedious task to perform. An effective Automated Speech Recognition system can accurately transcribe these calls, making it easier to search through call history for specific context and content with very less time and manual effort. With text transcription available calls can be automatically monitored, improving QoS through keyword search and sentiment analysis. ASR for Call Center requires more robustness as telephonic environment are generally noisy. Moreover, there are many low-resourced languages that are on verge of extinction which can be preserved with help of Automatic Speech Recognition Technology. Urdu is the 10th most widely spoken language in the world, with 231,295,440 worldwide still remains a resource constrained language in ASR. Regional call-center conversations operate in local language, with a mix of English numbers and technical terms generally causing a "code-switching" problem. Hence, this paper describes an implementation framework of a resource efficient Automatic Speech Recognition/ Speech to Text System in a noisy call-center environment using Chain Hybrid HMM and CNN-TDNN for Code-Switched Urdu Language. Using Hybrid HMM-DNN approach allowed us to utilize the advantages of Neural Network with less labelled data. Adding CNN with TDNN has shown to work better in noisy environment due to CNN's additional frequency dimension which captures extra information from noisy speech, thus improving accuracy. We collected data from various open sources and labelled some of the unlabelled data after analysing its general context and content from Urdu language as well as from commonly used words from other languages, primarily English and were able to achieve WER of 5.2% with noisy as well as clean environment in isolated words or numbers as well as in continuous spontaneous speech. Show less

Call Centers have huge amount of audio data which can be used for achieving valuable business insights. This requires transcription of phone calls which is manually tedious task to perform. An effective Automated Speech Recognition system can accurately transcribe these calls, making it easier to search through call history for specific context and content with very less time and manual effort. With text transcription available calls can be automatically monitored, improving QoS through keyword… Show more

Call Centers have huge amount of audio data which can be used for achieving valuable business insights. This requires transcription of phone calls which is manually tedious task to perform. An effective Automated Speech Recognition system can accurately transcribe these calls, making it easier to search through call history for specific context and content with very less time and manual effort. With text transcription available calls can be automatically monitored, improving QoS through keyword search and sentiment analysis. ASR for Call Center requires more robustness as telephonic environment are generally noisy. Moreover, there are many low-resourced languages that are on verge of extinction which can be preserved with help of Automatic Speech Recognition Technology. Urdu is the 10th most widely spoken language in the world, with 231,295,440 worldwide still remains a resource constrained language in ASR. Regional call-center conversations operate in local language, with a mix of English numbers and technical terms generally causing a "code-switching" problem. Hence, this paper describes an implementation framework of a resource efficient Automatic Speech Recognition/ Speech to Text System in a noisy call-center environment using Chain Hybrid HMM and CNN-TDNN for Code-Switched Urdu Language. Using Hybrid HMM-DNN approach allowed us to utilize the advantages of Neural Network with less labelled data. Adding CNN with TDNN has shown to work better in noisy environment due to CNN's additional frequency dimension which captures extra information from noisy speech, thus improving accuracy. We collected data from various open sources and labelled some of the unlabelled data after analysing its general context and content from Urdu language as well as from commonly used words from other languages, primarily English and were able to achieve WER of 5.2% with noisy as well as clean environment in isolated words or numbers as well as in continuous spontaneous speech. Show less

Speech Recognition Technology has evolved since the past few decades and has helped in various areas like telecommunication, Human-Computer Interaction, medical and security. Speech Recognition saw great improvement from 1970s onward with advent of Statistical HMM-GMM based systems but the accuracy problem always remained in noisy or production environment. The increase in computational powers through GPU allowed usage of Neural Networks but still Human Speech Recognition outperforms ASR in… Show more

Speech Recognition Technology has evolved since the past few decades and has helped in various areas like telecommunication, Human-Computer Interaction, medical and security. Speech Recognition saw great improvement from 1970s onward with advent of Statistical HMM-GMM based systems but the accuracy problem always remained in noisy or production environment. The increase in computational powers through GPU allowed usage of Neural Networks but still Human Speech Recognition outperforms ASR in terms of accuracy which, contrary to popular belief, means that Speech Recognition is not a solved problem. It is also not an AI-only problem that can be solved through a model-centric approach. It is a multi-faceted nuanced problem which involves speaker-based, linguistic-based, Dataset-based, environment \& recording, Security \& Privacy-based and finally Model Training (AI, Statistical or hybrid) and Deployment Challenges. Hence, a detailed survey is carried out to show the evolution of ASR and highlight all the challenges faced in Speech Recognition Technology which can help ASR researchers understand the overall bigger picture in this field. It provides more questions than answers to solve in ASR which will help ASR Researchers to work on to improve this field. Show less

Open for Business by Sage Khan & Rahim Ali explores the transformative role of open-source
software in the corporate landscape. It examines the evolution of software licensing, from the early days of
copyright and proprietary models, championed by figures like Bill Gates, to the rise of open-source principles
driven by the Free Software Foundation and the GNU Project. The book delves into how open-source projects,
supported by the Linux Foundation, have become integral to modern IT… Show more

Open for Business by Sage Khan & Rahim Ali explores the transformative role of open-source
software in the corporate landscape. It examines the evolution of software licensing, from the early days of
copyright and proprietary models, championed by figures like Bill Gates, to the rise of open-source principles
driven by the Free Software Foundation and the GNU Project. The book delves into how open-source projects,
supported by the Linux Foundation, have become integral to modern IT infrastructure, fostering innovation,
security, and cost-efficiency. It highlights the growing specialization in fields like DevOps, AI, and
cybersecurity, and discusses the challenges companies face in maintaining in-house expertise, often leading to
outsourcing for convenience. Khan critically assesses the impact of corporate intervention on open-source
projects, using case studies of Red Hat, HashiCorp, and others to illustrate how business decisions can
sometimes undermine the open-source ethos. Despite these challenges, the book emphasizes the resilience of
the open-source community through initiatives like project forks and alternative licensing models. ”Open for
Business” advocates for the adoption of open-source solutions to enhance organizational learning and flexibility
while balancing corporate needs for scalability and support. It calls for a renewed commitment to the
principles of transparency, collaboration, and freedom in the software industry.

Enroute Traditional Publishing. Will be published by end of 2024 Show less