Source: ATYUN AI platform
Facebook Voice computing revolution is a bit slow. It did not voice assistant, it's smart speaker is still under development, but some applications like Instagram are not fully equipped with audio communication. However, according to experiments found in the Facebook code, plus the new patent application, it may be subject to change.
It is reported that the upcoming Portal intelligent speakers designed for remote video chat with family and design, including the elderly and children may encounter problems mobile phone. Improved transcription and speech to text-to-speech conversion may be connected to the Messenger user input medium, and keep them in a chat application, and not biased in favor of SMS.
Given Facebook has been a step late, it needs to solve practical problems through a powerful utility. The following are the latest developments in Facebook in the field of voice and how it past experiments for its next major boost to lay the foundation.
Aloha
Facebook is the name of Aloha to develop their own voice recognition feature for Facebook and Messenger applications, as well as external hardware, might be that it is developing a video chat intelligent speaker. In the Android app and Facebook Messenger, a mobile researchers Jane Manchun Wong first saw the prototype Aloha user interface.
Highlights Click Original: Facebook's voice assistant Aloha suspected exposure
Aloha voice test, when the user speaks in a message thread, the horizontal blue bar expands and contracts, to visualize the amount of speech recognition and text transcription. The function code is described as the establishment of an external Wi-Fi or Bluetooth devices. The software can run on Facebook's hardware and software, similar to Google intelligent assistant that runs on your phone and Google Home speaker.
Years ago, Facebook interested in developing their own voice recognition software, designed for interaction among friends accurate transcription. And the way we deal with computer assistant formal Alexa or Google Home Amazon, etc. compared to those speech patterns are usually more casual, colloquial, quick and full of slang.
Code Wong also found the Aloha Facebook logo contains the image of the volcano. Someone saw a Facebook Aloha Setup bot on Facebook employee phone, above a similar logo.
If Facebook can solve this problem, it can provide its transcriptional function in Messenger and other parts of the site, so users can cross-media communication. It may make comments or send out messages to friends when you can not see the screen. Then, the recipient can read the text without having to listen to it like as a voice message. This function can also be used to power the voice navigation Facebook application, in order to achieve a better hands-free use.
Smart Speaker Portal
Facebook video chat intelligent speaker originally code-named Aloha, but later changed its name to Portal, Business Insider's Alex Heath and now Cheddar in August 2017 for the first time reported. Fearing exacerbate privacy scandal triggered by analyst firm Cambridge, it was postponed.
A new patent application display, Facebook is considering the establishment of an intelligent speaker, as early as December 26, 2016, applied for a patent for the device in the shape of a cube. This patent depicts a "speaker decorative design" by the Baback Elmieh, Alexandre Jais and John Proksch-Whaley invention. Facebook in September of that year acquired Elmieh of start-up companies Nascent Objects, Facebook and now he is the secret of Building 8 hardware laboratory technical project leader.
The long road to Voicebook
Facebook trimmed long speech in the field. By 2015, it acquired a start-up company Wit.ai natural language processing, the company runs a platform for developers to build voice interface, but then pushed into the Wit.ai Messenger platform team, focusing on bot. Facebook also started testing voice clips are automatically transcribed into text Messenger in 2015, which could be seen above the basic function of Aloha.
The next year, Facebook head of David Marcus claimed that "this is not what we are now actively working" but he added, "In a way, it is clear that as we develop more and more in Messenger function and interaction, we will begin processing voice switching and interfaces. "However, a source said Facebook's secret language voice technology group has been exploring opportunities. Facebook also began to play a sound instead of the video for the user who wants to test real-time audio capabilities.
2017, Facebook Pages provide automated captions for video and voice search capabilities are being developed. This year, Facebook began to try to voice clips as status updates and stories, these users may have difficulty typing in their mother tongue. But executives did not talk about voice initiatives.
Design head of Luke Woods most detailed comments from Facebook, he describes the TechCrunch Disrupt 2017 voice search, said it was "very promising."
It is unclear what will happen Facebook of Aloha. For intelligent speaker and Facebook applications, which can be an operating system or a voice interface and transcription functions. It may also be the same as the M to become a more mature voice assistant. Perhaps it could become a bridge Facebook and other voice ecosystem.
More voice recognition data set Click "here" to download
This switched ATYUN artificial intelligence media platforms, the original link: Facebook's voice assistant Aloha suspected exposure
more recommendations
Boston Dynamics robot army struck! Ten robot dog pulling trucks, to be put into commercial
Facebook AI may be the human voice in audio to convert between singer
