US intelligence unit launches $50k speech recognition competition

 

There’s a new high-tech competition afoot that challenges participants to design speech-recognition software that can decipher conversations and other speech that happens over microphones in noisy, echo-prone situations.

The $50,000 challenge comes from researchers at the Intelligence Advanced Research Projects Activity (IARPA), within the Office of the Director of National Intelligence. The completion, known as Automatic Speech recognition in Reverberant Environments (ASpIRE), hopes to get the industry, universities or other researchers to build automatic speech recognition technology that can handle a variety of acoustic environments and recording scenarios on natural conversational speech.
The ASpIRE challenge seeks to inspire the development of innovative speech recognition systems that can be trained on conversational telephone speech, yet work well on far-field microphone data from noisy, reverberant rooms. Challenge competitors will be given access to sample data against which they can test their algorithms that are different from the test set, but provide a good representation of microphone recordings in real rooms, IARPA stated.
Contestants are then asked to use their algorithms to submit transcriptions for the test set. The algorithm that produces the lowest word error rate in the single microphone condition will receive $30,000, and the algorithm that produces the lowest word error rate in the multiple microphone condition will receive $20,000. The single microphone condition evaluation will run February 4-11, 2015. The multiple microphone condition evaluation will run February 12-19, 2015.
There are two evaluation conditions, according to IARPA:

• The Single Microphone (single-mic) Condition tests the ability to mitigate noise and reverberation given a single microphone recording (selected randomly) from speech recorded in several rooms with a variety of microphones. Single-mic evaluation data will be made available at 10:00 A.M. Eastern Standard Time 04-Feb-2015 and submissions must be received by 11:59 P.M. Eastern Standard Time 11-Feb-2015 to be eligible for award.
• The Multiple Microphone (multi-mic) Condition tests the ability to mitigate noise and reverberation given all of the microphone recordings of speech recorded in several rooms with a variety of microphones. Multi-mic evaluation data will be made available at 10:00 A.M. Eastern Standard Time 12-Feb-2015 and submissions must be received by 11:59 P.M. Eastern Standard Time 19-Feb-2015 to be eligible for award.
• In both conditions, word error rate will be used as the objective measure of performance. Contestants can participate in either or both conditions, IARPA stated.