Amazon has filed a patent application with the US Patent and Trademark Office, which describes a technology that allows Echo and other Alexa-enabled devices to intercept what you say before a wakeword that "Alexa" pronounced. Currently, Alexa devices only detect and transmit audio to Amazon servers if a wakeword is detected. If Amazon decides to develop or implement the technology, an Alexa-enabled device will constantly detect and delete what you say using the device's local memory storage.
The patent application that was published today provides insight into Amazon's ambitions to extend the ability to speak speech recognition technology. Alexa devices cannot currently understand commands where wakeword comes after or is in the middle of a sentence. But images in the patent application offer " Play some music, Alexa " and " Play some music, Alexa. The Beatles, please " as examples. "
" While such formulations may be natural to a user, the current speech processing systems are not configured to handle non-wakeword commands, "patent application authors Kurt Wesley Piersol and Gabriel Beddingfield wrote." Offered is a system to address this problem. "[1
According to the patent application for a wake word is detected, Alexa can "look backwards" to determine if the command came before the wakeword and use pauses in speech to identify the beginning of the command. At that time, the sound would be sent from the device to Amazon servers for full processing. The idea is similar to Apple's Live Photos technology, which captures 1.5 seconds before and after camera shutter is pressed.
This process, the patent application states, is designed so that all captured numbers do not have to be sent to Amazon, "Thus, addressing privacy concerns associated with an" always-on "voice processing system. In addition, the device may be configured to store only 10 to 30 seconds of audio at a time "to avoid recording too much speech and causing privacy concerns."
Still, the potential technology means that Alexa devices will take up more conversations and more sound, including speech, as it seems commands. While Amazon's speech recognition software is constantly improving, Echo and third-party Alexa devices accidentally record lots of audio extracts triggered by false-wake words. This could expose several customer recordings to Amazon employees and contracts to review and annotate audio files to enhance Alexa's technology as well as others. In May 2018, an Amazon Echo accidentally shared parts of a woman's private conversation as a voice message to one of her husband's employees. It is also unclear whether customers will be able to turn off pre-wakeword recording.
Privacy Protection is the patent application just a part of Amazon's attempt to make voting more natural and human. But in order for Amazon to do so, it seems that we should let Alexa listen to several of our conversations.
You can review and delete voice recordings captured by Echo and third-party Alexa devices by going to Settings > Alexa Account > Alexa Privacy in Alexa- the app or https://www.amazon.com/alexaprivacy.