#20.1 Speech Synthesis Workshop 2025 with Xin Wang

Join us as we sit down with Hynek and Morgan to talk all things speech research. We’re getting into the real stuff: why they started, whether they actually had a "grand plan" or were just following their curiosity, and how they decide which projects are actually worth the effort.We also get into the technical weeds (in a good way) on things like energy efficiency and the battle between neural nets and knowledge-driven models. Plus, there’s plenty of solid advice for students on finding mentors, dealing with those brutal paper rejections, and knowing when to move fast versus taking it slow.0:00:00 - intro00:00:11 - introduction of the guests00:01:16 - Hynek00:02:02 - Morgan00:04:08 - research focus: why did they get into speech research?00:08:46 - did they have a plan on how to influence the field?00:11:29 - were they able to fake that they had a grand goal when they were rather driven by curiosity?00:19:13 - how do they chose the projects to pursue?00:24:09 - what things did they pick up from working and collaborating with other people?00:27:32 - what interdisciplinary knowledge is so important right now to drive the next big innovation forward?00:34:36 - how to improve current neural nets; end-to-end vs. knowledge-driven approaches; energy efficiency00:48:53 - mentoring: how can students approach seniors?00:56:02 - moving fast and breaking things vs. moving slowly00:58:04 - mentoring students01:08:24 - how to deal with reviewers and rejections?01:15:40 - what are they doing nowadays?01:25:08 - question from the previous guests: pivotal moment in your career01:31:15 - question for the next guest: will better deep neural networks solve everything speech related? How can we convince students to write papers on why they did their research?01:32:35 - farewell01:33:10 - outro

Jhansi shares her unique dual perspective—firstly as a researcher pushing boundaries in the field, and secondly as a member of the PhD panel, where she helped guide the next generation of scientists. Host: Spyretta LeivaditiPost-Production: Mohammed Mosuily, Pascal Hecker, Spyretta Leivaditi

Yuanyuan joins us to talk about her research in inclusive speech technology and the power of representation in AI. Whether she’s collecting personalized audio-video datasets or organizing the next big Interspeech, Yan is a champion for community.Tune in for a candid conversation about:Diverse Data: Accented, child, and disaster-related speech.Leadership: Transitioning from workshop participant to organizer.The PhD Mindset: Overcoming stress and prioritizing mental health.Host: Spyretta LeivaditiPost-Production: Mohammed Mosuily, Pascal Hecker

Fresh from the Young Female Researchers in Speech Workshop (YFRSW) 2025, Chin-Jou Li joins us to share her experience as a rising voice in the field. Chin-Jou discusses her poster presentation "Towards Inclusive ASR: Investigating Voice Conversion for Dysarthric Speech Recognition in Low-Resource Languages", her reflections on the mentoring and community at the workshop.Host: Spyretta LeiavditiPost-Production: Mohammed Mosuily, Pascal Hecker

Julian Zaidi and Marc-André Carbonneau are discussing the practical differences between academic speech research and industrial application within the video game sector. The guests compare the Speech Synthesis Workshop to the broader Interspeech conference, noting that smaller venues allow for more technical discussions on voice cloning and evaluation metrics. They highlight critical challenges for the industry, such as maintaining speaker identity across varied emotions and the necessity for compact models that run efficiently on local hardware. The conversation further explores the cultural shift required when moving to a commercial environment, where consistent reliability and effective communication with non-experts are prioritised over theoretical novelty. Ultimately, the speakers emphasise that while academia thrives on exploration, industry demands functional, high-performance tools that meet the strict expectations of players.Host: Paige TuttösiPost-Production: Zhengjun Yue, Pascal Hecker

We travel to the Speech Synthesis Workshop (SSW) 2025 to sit down with Xin Wang, Project Associate Professor at the National Institute of Informatics (NII) in Japan. Xin shares insights into his groundbreaking work. We also explore the highlights of this year's workshop and the evolving landscape of synthetic media.Host: Paige TuttösiPost-Production: Tianyi Zhao, Pascal Hecker

Kick off the Speech Synthesis Workshop 2026 with an opening message from Matt Coler. In this introductory segment, Matt sets the stage for a year of breakthrough research, collaborative innovation, and the evolving landscape of synthetic speech. Whether you’re a veteran researcher or a curious newcomer, the journey starts here.Host: Matt ColerPost-Production: Zhengjun Yue, Pascal Hecker

From strangers to champions! 🏆 Meet Yassine Tijani and Joas Walraven, two young innovators who crossed paths at the Speech Science Festival and decided to team up on the spot. Tune in to hear how they dominated the Hackathon, what sparked their interest in speech tech, and the story behind their winning collaboration.Host: Pascal HeckerPost-Production: Pascal Hecker

Live from the Speech Science Festival! 🎤 We sat down with the University of Waterloo’s Nada Gohider and her wonderful children to get an inside look at this year's festivities. From hands-on events to the science of sound, they’re sharing their favorite moments and the unique energy that makes this festival a must-attend for families and researchers alike.Host: Pascal HeckerPost-Production: Pascal Hecker

Your memories, composed. 🎹 Paul Raingeard de la Blétière is bridging the gap between technology and nostalgia. This demo showcases his innovative robot that listens to life stories and transforms them into fully-realized songs. From chosen instruments to lyric-matching memories, it’s a personalized symphony for the soul, specifically crafted for the dementia community.Host: Janice HuangPost-Production: Wei Xue, Pascal Hecker