SpeD Conference - Speech Technology and Human-Computer Dialogue

Great news! 🎉
The SPED 2025 Proceedings have officially been indexed in IEEE Xplore!
Proud of all the contributors and grateful to everyone who supported the conference.

You can access the proceedings here:
👉 https://ieeexplore.ieee.org/xpl/conhome/11251505/proceeding

SpeD 2025 has come to an end — and what an inspiring few days it has been!

This year’s edition, organized by Universitatea POLITEHNICA din București and Universitatea Tehnica din Cluj-Napoca, brought together 54 participants from 8 countries — researchers, students, and industry experts passionate about speech and dialogue technologies.

Across 6 lively sessions — Audio Deepfakes and Forensics, Natural Language Processing, Emotion, Speaker Recognition, Diarization & Multimodal, Synthetic Speech, Automatic Speech Recognition, and Signal Processing — we shared exciting new research, exchanged ideas, and built new connections.

We were honored to host three outstanding keynote speakers:
👉 Tomi Kinnunen, Cassia Valentini, and Andrei Paun — thank you for your inspiring and thought-provoking talks!

A big thank you to our amazing partners Infineon Technologies AG Romania and Orange Romania, and their representatives Valentina Davidoiu and Razvan Mihai for their support and collaboration in making this event possible.

We also thank our technical sponsors IEEE and EURASIP for their continued endorsement and trust.

📸 Scroll through for some highlights from the conference — including our group photo and moments from the sessions.

It’s been wonderful to see the community grow and share knowledge once again.

What was your favorite takeaway from SpeD 2025? 💬
See you all at the next edition! 💫

That’s a wrap on Day 2 and the end of the SpeD2025 Conference! 🎉

Over the past two days, we’ve explored cutting-edge research, discussed the future of speech and AI technologies, and connected with an inspiring community of experts, innovators, and partners.

A heartfelt thank-you to all our speakers, participants, and sponsors — your insights and energy made this conference a true success.

The conversation continues tomorrow with the Language Data Space Country Workshop – Romania, where we’ll dive deeper into European initiatives on language data, AI, and collaboration.

See you there! 🇷🇴

This afternoon at the SpeD2025 Conference, we’re proud to highlight our valued partners who are driving innovation through research and collaboration.

Orange Romania, through Răzvan Mihai, presented “Orange Romania Research and Development Ecosystem”, showcasing how the company fosters cutting-edge advancements and supports the local tech community.

Infineon Romania, through Valentina Davidoiu, followed with “Infineon Romania – Semiconductors R&D Competence Center,” offering a deep look into their pioneering work in semiconductor technology and research excellence.

A huge thank-you to both Orange Romania and Infineon Romania for their continuous support and commitment to innovation.

We’re excited to continue Day 2 of the SpeD2025 Conference with our last keynote:

“Evaluating Speech Technology – Challenges and Opportunities”, presented by Dr. Cassia Valentini-Botinhao from the University of Edinburgh.

This session explores the critical question of how we measure progress in speech technology — balancing technical performance, human perception, and ethical considerations.

Good morning from the SpeD2025 Conference — Day 2! ☀️

After an inspiring first day filled with innovation, collaboration, and forward-thinking ideas, we’re ready to dive into another day of impactful discussions and knowledge sharing.

That’s a wrap on Day 1 of the SpeD2025 Conference! 🎉

What an incredible start — filled with inspiring keynotes, thought-provoking discussions, and valuable connections.

Highlights included:
🔹 Professor Tomi Kinnunen (University of Eastern Finland) on building trust in the age of speech deepfakes
🔹 Paul-Andrei Păun (Romanian Academy) sharing new research directions in AI at the ICIA “Mihai Drăgănescu”

A huge Thank you! to our speakers, participants, and organizers for setting such a high bar.

We’re looking forward to more learning, innovation, and collaboration in the days ahead!

We’re continuing an exciting first day at the SpeD Conference with our second keynote:

“New Research Directions of the Romanian Academy Research Institute for Artificial Intelligence ‘Mihai Drăgănescu’ (ICIA)”, presented by Paul-Andrei Păun of the Romanian Academy.

This session offers an inside look at cutting-edge AI research emerging from ICIA — exploring new frontiers in artificial intelligence and the visionary work shaping Romania’s contribution to global innovation.

We’re excited to continue the SpeD2025 Conference with our first keynote:

“Building Trust in the Age of Speech Deepfakes: From Detection to Source Tracing and Robust Biometrics”

Presented by Professor Tomi H. Kinnunen from the University of Eastern Finland, this session explores one of today’s most critical challenges — how to ensure authenticity and trust in a world of rapidly advancing AI-generated speech.

And we’re off! The SpeD2025 Conference has officially begun.

Excited to spend the next few days engaging in meaningful discussions, sharing best practices, and celebrating innovation in special education.

We’re excited to announce that we’re all set and ready to kick off the SpeD2025 Conference!

Looking forward to an inspiring and impactful event ahead. Let’s get started! 🚀

Exciting news! The Conference Schedule is now live.
Check out the full program and get ready for inspiring sessions, insightful talks and valuable networking opportunities.

🔗Explore the schedule here: https://sped.pub.ro/conference-program/

We look forward to seeing you at !

Conference Program - SpeD 2025 Provisional Conference Schedule Time (Mon) Monday (October 20) Time (Tue) Tuesday (October 21) 9:00 Registration. Opening 9:00 Registration 9:30 KeynoteTomi Kinnunen 9:30 KeynoteCassia Valentini‑Botinhao 10:30 Coffee break 10:30 Coffee break 11:00 Oral Session: Audio deepfakes and forensics 11:00 ...