Amazon unveiled the fourth generation version of its key Echo smart speaker, bringing a new spherical design and improved sound efficiency. But the biggest improvement is a modern, on-device speech recognition module that can process your Echo audio locally, making your requests quicker than ever before.

Another major addition to the latest Echo is what Amazon is calling the AZ1 Neural Edge Silicon Module, which will process the audio of your voice requests using local computer speech recognition algorithms before sending the command to the cloud. The method promises to save your Echo hundreds of milliseconds in response time.