More of the mobile world is coming around to seeing (hearing?) that one of the big problems with speech-enabled mobile applications has been failure to work in noisy environments. Hence the need for YANCS (or Yet Another Noise Cancellation Solution) like Audience, Inc., and Ditech Networks

The latest news on this front is a $15 million investment in Audience, Inc. as the fourth round in funding of the nine year old company. Investors, who include New Enterprise Associates, Tallwood Venture Capital and VentureTech Alliance see the value in Audience’s area of specialization, which it calls “subjective noise suppression.” According to company claims, its custom chips help phones or automotive electronics “hear” more like humans by eliminating noise, such as airport PA systems, restaurant noise, car noise, or music and background conversations.” No small task! And one that is crucial to endearing mobile customers to using speech-enabled services offered through voice portals or automated self-service applications.

Audience, Inc., was founded in 2000 and has marketed its processors primarily in the Asia/Pacific region. The products are offered as custom DSP chipsets. According to the company’s Web site, they have been incorporated into handsets from Sharp, LG and Pantech (the latter two being marketed by South Korea’s SKTelecom) and its solutions have also been deployed in Japan, “powering” DoCoMo handsets.

One of its primary rivals is Ditech Networks, which was one of the darlings of the recent eComm Conference in Burlingame. As we noted in this post, Ditech noise cancellation is a crucial part of toktok, which uses a unique phrase in the course of a phone conversation to toggle over to speech-enabled applications such as calendar management or updating entries to Facebook.

The investment should be money well spent. It is important to promote accurate recognition over wireless networks, including Bluetooth, and in the face of challenges posed by employing tiny microphones commonly employed in wireless earpieces or headsets. Perfection is probably not possible, but acceptance of voice hinges on minimizing those moments when applications go awry in noisy environments.

Categories: Articles