Talking cars closer to reality with voice biometrics products from Aculab, Kardome
Voice biometrics companies are eager to get drivers talking to their cars. And with the rise of generative artificial intelligence, they are getting better at turning cars into great listeners.
Speech recognition company Aculab has presented a new access control solution based on biometrics, which allows users to open vehicle doors using voice commands.
The product comes from a collaboration between the UK-based company and two German firms: conversational technologies and generative AI company Frank Reply, a subsidiary of the Reply Group, and automotive supplier and manufacturer Brose. Called VoiceSentry, it can be used to replace traditional car keys, preventing theft and making it easy for a recognized user to open the vehicle. According to a release, it analyzes thousands of unique voice features within spoken commands, providing high accuracy when identifying vehicle owners.
Aside from VoiceSentry, Aculab developed the FaiSentry face identification system last year, as well as SentryFusion, a multimodal biometric system combining voice and face biometrics. The company sells its tools, which include voice authentication and Communications Platform as a Service (CPaas), to over 80 countries globally.
Kardome launches in-car multi-speaker speech, voice recognition
Israeli startup Kardome has introduced a new speech and voice recognition product, which it claims is the only one on the market that can individually capture and biometrically identify up to six speakers in three seat rows with a single mic array.
The company says that its product Kardome Mobility is unique because of its 3D spot-forming mode. The spot-forming-based Audio Front End (AFE) uses reverberations to separate sounds from different locations, allowing it to recognize speech in noisy environments with multiple speakers. It includes multichannel acoustic echo cancellation, noise reduction, source speakers separation and speech localization.
Competing systems, on the other hand, use beamforming techniques that are limited when separating multiple speakers. To overcome the limitations of beamforming algorithms, manufacturers deploy microphones in the roof liner of the car in four to six locations. This results in additional weight, installation costs, and the added bill of materials costs, says the company in a release.
“Kardome Mobility changes the paradigm that listening to all people in a car with a single sensor is an impossible mission,” says Kardome CEO Dani Cherkassky. “We are proud to equip cars with human-level ‘listening capabilities’ that enable a reliable voice user interface as well as intelligible hands-free communication, all while minimizing the cost of the system to manufacturers.”
The audio software company showcased its speech recognition tech during this year’s Paris VivaTech conference in June, integrated into a concept car called H1st Vision made by Software République, a European collaborative project focused on mobility and supported by the Renault Group.
This year in March, Kardome received a US$1 million investment from Japan’s Automotive Fund, a subsidiary of the automotive industry portal Marklines.