Base Styles/Icons/Pin/Default Created with Sketch. Base Styles/Icons/phone Created with Sketch. Base Styles/Icons/Some/Linkedin/Default Created with Sketch. Base Styles/Icons/Menu/Burger Default Created with Sketch. Base Styles/Icons/Close/Default Created with Sketch. Base Styles/Icons/lock/open Created with Sketch.

Speech recognition

Our goal is to generally improve the speech recognition methodology with the help of the new algorithms developed in Aalto University. Speech recognition offers challenging benchmarking tasks for efficient algorithms that can process and learn to represent large quantities of data. In addition to improving the acoustic models of phonemes we aim at developing new learning statistical language models for difficult large vocabulary continuous speech recognition tasks. The relevant pilot applications in our group range from unlimited vocabulary continuous dictation in different languages to audio indexing, adaptive speech synthesis, and speech-to-speech translation.
Speech_Recognition_Aalto_University

At the national Center of Excellence in Computational Inference Research the research group belongs to the flagship of Intelligent Information Access.  In language modeling, speech translation, video indexing and speech synthesis we work closely together with the Adaptive Natural Language Processing, Computational Cognitive Systems, Content-Based Image and Information Retrieval, and Speech Analysis research groups.

The research group is led by Professor Mikko Kurimo.

Research Overview

Speech recognition offers challenging benchmarking tasks for efficient algorithms that can process and learn to represent large quantities of data. In addition to improving the acoustic models of phonemes we aim at developing new learning statistical language models for difficult large vocabulary continuous speech recognition tasks.

Automatic speech recognition (ASR) and the modern parametric speech synthesis (text-to-speech, TTS) systems typically share the same underlying statistical modeling scheme, the hidden Markov models (HMM). The acoustic features and their statistical models (HMMs) in ASR and TTS resemble to each other. The language modelling tasks in ASR and TTS differ more, but some shared tools such as deep neural networks and unsupervised and semi-supervised morphological analysis can be helpful for text analysis in both. Speech-to-speech translation (S2ST) is a challenging application that combines both these problems with statistical machine translation (SMT).