The 2nd Dutch Speech Tech Day kicked off yesterday to a sold-out crowd of professionals, scientists, and students from all facets of speech technology in The Netherlands. Hosted at a building which can be described as a beautiful cube in Hilversum, Beeld en Geluid, the inspiring venue seemed a great match for the content.



Organized by the Open Speech Technology Foundation (Stichting Open Spraaktechnologie), Dr Odette Scharenborg (TU/Delft) opened the event which started with a keynote by Dr Olya Kudina (also TU/Delft) before dedicated sessions on industry and academia. There was also a poster session featuring student work and an information market (special shout out to SpeakSee whose demo showed the possibility of near simultaneous translation for Mandarin, Cantonese and English featuring MSc Voice Tech students).
In the academic session, I gave a talk entitled “Fostering Diversity Through Speech Technology” which highlighted contributions by MSc Voice Tech grads Dragoș Alexandru Bălan, Golshid Shekoufandeh, and Spyretta Leivaditi alongside highlighting the contributions of the students who worked on 9292hoi: Liu, Ouyang, Lin, Wei, Hongell, and Ho — together with ReadSpeaker. Student work fuels innovation!
As part of the talk, I polled the audience on challenges they faced with ASR . The responses appear below. Unsurprisingly everyone faced some difficulties with speech recognition, particularly misrecognizing names or common words and struggling with identifying the speech signal in a noisy environment. The point is that these issues are not ephemeral, complex issues for the far future, but are actually manageable topics that academic research can address through graduate research.
I concluded with a reflection on how to leverage this moment in the development of the field to promote values like inclusivity and accessibility for speech technology so that the future of the field may be impactful, delightful, and profound.
All in all: Great event full of familiar faces, newcomers, and more proof that The Netherlands is a hotbed of speech tech innovation.