On May 4, 2026, the IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2026) officially opened in Barcelona, Spain.
Researchers and industry professionals from around the world gathered in the city for several days of discussions, presentations, and exchanges across signal processing, speech, acoustics, and AI-related fields.
One notable detail this year is the conference acceptance rate, which dropped to around 40%, compared with nearly 50% in previous years.
The lower acceptance rate reflects both the growing volume and quality of submissions, as well as increasingly competitive review standards — making accepted papers at ICASSP 2026 even more selective than before.
Among all accepted papers at ICASSP 2026, submissions from China accounted for approximately 65% of the total accepted papers, reflecting the country’s strong presence in the global AI research community.
Looking at the distribution by research area, Machine Learning for Signal Processing (MIL) represented the largest share at 24%, followed by Speech and Language Processing (SL) at 16%. The results suggest that AI research continues to move toward more dynamic, sequence- and interaction-oriented modeling.
- SAM (Sensor Array and Multichannel Signal Processing): 2%
- AS (Audio and Acoustic Signal Processing): 4%
- SL (Speech and Language Processing): 16%
- AA (Audio Analysis and Enhancement): 8%
- IF (Image and Video Processing): 5%
- MMI (Multimedia and Human-Machine Interaction): 8%
- SPCOM (Signal Processing for Communications): 3%
- TM (Time-Frequency Analysis and Transform): 5%
- MIL (Machine Learning for Signal Processing): 24%
- Others: 12%
In the speech-related tracks alone, a total of 898 papers were accepted, with speech recognition, speech enhancement, and speech synthesis remaining among the most active research areas this year.
For more detailed statistics and a full paper overview, please visit:
https://nanless.github.io/audio-paper-digest-blog/posts/icassp2026-summary/
To give attendees a taste of the local culture and atmosphere of Barcelona, the ICASSP 2026 organizing committee prepared a lively cultural performance featuring flamenco, traditional music, and other local artistic elements — adding a vibrant Mediterranean touch to the conference.
On the first day of the conference, Magic Data and its open-source community platform MagicHub welcomed attendees at the exhibition area, where the team shared and discussed multilingual natural conversational datasets with researchers and industry professionals from around the world.
The booth stayed busy throughout the day, with many visitors stopping by to exchange ideas on conversational AI, speech technologies, and open data initiatives.
The specially designed tote bags also became unexpectedly popular at the venue, with many attendees carrying them around the conference shortly after visiting the booth.
Great first day at ICASSP 2026 — lots of insightful discussions and meaningful exchanges around conversational AI and speech technologies.
Excited for the next few days and the opportunity to connect with more researchers, partners, and industry teams from around the world.
