Scribd
Upload
32 views11 pages

Automatic Speech Recognition For Indian Languages: Comprehension and Analysis

The document discusses the technical challenges of automatic speech recognition for Indian languages including their diversity in phonetics and pronunciation. It covers the key components of ASR systems, research and development efforts, and issues faced including a lack of standardized speech and text corpora. The conclusion emphasizes that ASR models need adjustment to understand cultural and linguistic differences to improve accuracy and access to technology.

Uploaded by

sayalibarhate2717
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
32 views11 pages

Automatic Speech Recognition For Indian Languages: Comprehension and Analysis

The document discusses the technical challenges of automatic speech recognition for Indian languages including their diversity in phonetics and pronunciation. It covers the key components of ASR systems, research and development efforts, and issues faced including a lack of standardized speech and text corpora. The conclusion emphasizes that ASR models need adjustment to understand cultural and linguistic differences to improve accuracy and access to technology.

Uploaded by

sayalibarhate2717
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 11

Automatic Speech Recognition for Indian

Languages: Comprehension and Analysis


M.Sc. Computer Science

Presented By :
Rugveda Kushare
Shreya Ghadage
Introduction

◆ Technical intricacies of ASR


◆ Specific challenges in transliterating Indian languages
◆ Ongoing efforts to overcome these challenges
◆ Future directions for advancing ASR technology
◆ Better serving the linguistic diversity of India
OVERVIEW OF AUTOMATIC
SPEECH RECOGNITION

● Definition and Functionality of ASR


● Importance of ASR for Indian Languages
● Key Components of ASR Systems
DIVERSITY IN INDIAN LANGUAGES
• Variety of Languages and Dialects
• Unique Phonetics and Pronunciation
• Impact on ASR Accuracy
Word Error Rate
Word Error Rate (WER) is a metric used to evaluate the accuracy of an Automatic Speech
Recognition (ASR) system's transcription output.

Mathematically, the formula for WER is,


WER=((S+D+I)/N)*100

• 1. S is the number of substitutions (words in the reference that are replaced by incorrect words
in the recognized text),
• 2. D is the number of deletions (words missing in the recognized text compared to the
reference text),
• 3. I is the number of insertions (extra words present in the recognized text compared to the
reference text), and
• 4. N is the total number of words in the reference text
RESEARCH AND DEVELOPMENT IN
ASR
• Current State of ASR for Indian Languages
• Key Research Initiatives and Projects
• Challenges in Data Collection and Annotation
DATA PROCESSING

1. Audio Processing
2. Preprocessing and Text Cleaning
3. Reduction and Transliteration of Text
PROCESS OF TRANSLITERATION
• Transliteration of Indic Words
• Word transliteration from the English
Dictionary
• Reduction
ISSUES FACED IN ASR FOR INDIAN
LANGUAGES
• Linguistic, speaker, and channel variability pose challenges for ASR
engines.
• ASR systems must adapt to unpredictable speech signals for accurate
transcription.
• The lack of standard speech and text corpus presents challenges for Indian
languages.
• Gathering a corpus of speech requires careful attention and data extraction.
• Speech variables impact the effectiveness of ASR systems in various
applications
CONCLUSION
The research shows that making ASR work well for languages like Hindi,
Tamil, and others is tricky because they have different ways of speaking
and cultural meanings. We need to adjust ASR models to understand these
differences better. By doing this, we can improve how accurately ASR
understands and writes down what people say. ASR is important because it
helps more people access technology and keeps languages alive. We need to
keep learning and improving ASR to make it work well for all languages.
THANK YOU

You might also like