Scmlgroup

🧠✨ Webinar Alert! ✨🧠

Title: Spoofing Robust Automatic Speaker Verification (SASV)

🎙️ Speaker: Prof. Itshak Lapidot
👨‍🏫 Faculty, Afeka Academic College of Engineering, Israel
📨 Organized by: SCML Group
📧 Email: [email protected]

📅 Date: 25th September 2025 (Thursday)
⏰ Time: 15:00 CET (06:30 PM IST)
📍 Mode: Online (MS-Teams)

Registration Link: https://forms.gle/9s2oH8hueKCcXthj9

📅 Registration Deadline: 24th September 2025 (06:30 PM IST)

Abstract of the talk: Speaker recognition technologies evolve much in the resent years. As such it is not enough only to verify whether to speech segments belong to the same target speaker. There are additional challenges like speaker diarization, speaker clustering or speaker detection and more. In this work a problem of anti-spoofing is presented/spoofing robust speaker verification (SASV).Most anti-spoofing systems based on some transformation of the frequency domain, such as spectrograms, cepstrograms, etc. It will be shown that time domain carries an important information as well. Several time-domain based systems will be presented. It will be shown that time-domain information can be complimentary to the frequency-domain based countermeasure systems. Time-based hand-crafted embeddings also have a very important property explainability, which is of high importance in many domains, including forensics. At the end a new SASV system architecture based on ECAPA-TDNN and AASIST will be presented as well.

Upcoming Webinar Announcement:
We are pleased to invite you to the 19th webinar in our series, titled:
"Speech Assessments: An Overview"
Speaker:
Dr. Sneha Raman
Research Scientist
Department of Electrical Engineering
Indian Institute of Technology Bombay
Date: 24th May 2025
Time: 3:00 PM – 4:00 PM (IST)
Registration Link: https://forms.gle/vf89s61cYzp1iU8o6
Deadline of Registration: 23rd May 2025 (4:00 PM)

Join us for an insightful session where Dr. Sneha Raman will provide an overview of various approaches and methodologies in speech assessment, their applications, and emerging research directions.

Abstract of the webinar:
The intent of producing speech is varied – communicating with loved ones and strangers, interacting with service providers, reading aloud, public discourse, teaching and so on. In many scenarios, the produced speech is not intelligible or natural sounding, such as with non-native speech, disordered speech, children’s speech or TTS outputs. On the other hand, speech is perceived to be unintelligible or unnatural by hearing impaired individuals and in noisy environments. The speech science community, whether it is speech scientists, speech pathologists or machine learning experts, put in a lot of effort in making speech easier to comprehend and listen to, via interventions, speech enhancement algorithms and other methods. One common need in all these fields of research is to conduct appropriate, effective, accurate, use-case centric speech assessment methods.
Is a speech sample good enough for a normal hearing listener? Will it be understood by a hearing-impaired listener? How much has a non-native speaker improved after going for diction classes for three months? Is a child reading better at the end of the academic year compared to the start? All these are questions that can be answered by various speech assessment methods. In this talk I will give an overview of the several speech assessment techniques used today, from Likert scales to measures from brain signals. Then I will focus on two applications that I have worked on – disordered speech and oral reading fluency - where speech assessment is at the centrestage.

19th Webinar of the SCML Group

The topic of the webinar: Evolution of Speech Foundation Models
Invited Speaker: Dr. Nirmesh Shah, Senior Research Scientist, Sony Research India (SRI)
Date of the webinar: Sep 23, 2024
Time: 06:15 PM (IST) onwards
VENUE: online
Registration Link: https://forms.gle/a1nM7NpW5DaPmp6C9
Deadline of registration: Sep 22, 2024 (11:59 AM)

18th webinar of the SCML Group

Invited Speaker : Prof. (Dr.) Lipika Dey

Prof. (Dr.) Lipika Dey is currently a Professor of Computer Science at Ashoka University (Sonipat, Haryana) and Adjunct faculty at IIT Jodhpur. She also served as the Chief Scientist at TCS Research.

Registration Link: https://forms.gle/xb9NvzY9GvyhvW4t7

Date of the webinar : 23 Aug 2024 (Friday)
Time: 5:00 PM IST onwards (tentative)
Last date of Registration: 21 Aug 2024 (10 AM IST)
VENUE: online (joining link will be shared 1/2 days before the webinar)

Email-id: [email protected]

17th webinar of the SCML Group

Invited speaker: Prof. (Dr.) Pushpak Bhattacharyya (Professor, IIT Bombay, CSE Dept)

Abstract:
Natural Language Processing has undergone a huge transformation, transforming AI and Society. In this talk, we will present the foundations underlying this transformation along with cutting-edge applications. We will cover our own research and the insights gained therefrom.

Broader topic of the webinar : "Multilinguality, Multimodality, LLM and GenAI"

Registration Link: https://lnkd.in/grZ7Cs5i

Details:
Date: 1st July (Monday), 2024
Time: 5:30 PM (IST) onwards
Deadline of the Registration: June 30, 2024 (till 3:00 PM IST)