Qualcomm’s new smart speaker chipset promises better voice recognition, better audio, more

Ben Schoon | Mar 19 2019 - 7:45 am PT

As the smart speaker market continues to heat up, Qualcomm is jumping into the ring to provide a new chipset for OEMs to use. The new Qualcomm QSC400 makes some big promises for future products, here’s what you need to know.

Detailed in a press release today, the QSC400 is the first series of system-on-chip options from the company designed directly for smart speakers such as Google Home or the Amazon Echo. Qualcomm says that the new chips are designed with a few sore points of current speakers in mind, but voice recognition is the most interesting.

Apparently, this new chip can stay on standby for up to 25 times longer and can also pick up voice commands even when loud music is playing from the speaker itself. Anyone who uses their smart speaker for music can probably relate to having to shout to get the attention of the speaker. With these new chips, future speakers will apparently work better in that scenario, as well as from a further distance.

Utilizing our far field voice, multi-channel echo cancellation, and high performance, low-power multi-keyword detection algorithms (Qualcomm® Voice Assist), products built on the QCS400 SoCs can capture voice commands even during high-volume audio playback or from further afield.

Another notable strength of these new chips is audio quality. While it won’t turn a low-tier speaker like the Google Home Mini into a professional speaker by any means, the new single-chip amplifier should lead to louder sound overall. Plus, the chip can handle 32-channel Dolby Atmos and DTS:X. Notably, the QSC400 series also includes a chip designed for soundbars.

Qualcomm also says that these new chips should lead to shorter development cycles for manufacturers, better battery life on portable smart speakers, more powerful on-device AI, and more.

For users of smart speakers, the QCS400 SoCs support a more robust voice assistant experience with faster, smarter voice UI, even in noisy environments compared to our previous SoCs– including AI-based local automatic speech recognition, low-power, multi-keyword far-field voice pickup with beam-forming and echo cancellation, and support for cloud-based voice assistants.

Currently, it’s unclear exactly when we can expect any of the QCS400 series chips to debut in consumer products.