W3C home > Mailing lists > Public > public-wai-cc@w3.org > February 2021

RE: Clarifying speech recognition vs. voice recognition in Style Guide

From: Bakken, Brent <Brent.Bakken@Pearson.com>
Date: Wed, 3 Feb 2021 13:59:03 +0000
To: Steve Lee <stevelee@w3.org>, "public-wai-cc@w3.org" <public-wai-cc@w3.org>
Message-ID: <BYAPR07MB4198B4793CE2E85D56D96D1288B49@BYAPR07MB4198.namprd07.prod.outlook.com>
Just an FYI, Google Home uses both. We are a family of three and at any given time one of us can say "Hey Google, what is my name?" and it will answer with the correct name. It is using voice recognition all the time to determine who in the family is speaking, and obviously speech recognition to determine what we are saying. 

When a guest comes to our house and says "Hey Google, what is my name?" it will say something like "I don't know who you are." or "I don't remember who you are." Then they can say, "Hey Google, my name is John." It will say, "Okay, I will call you John." and it remembers their voice for future conversation.

Unfortunately, everything we asked about and commented on using google home ended up turning up as adds and spam email in all of my accounts because the device was registered in my name. So I reset it, unplugged it and sold it. It is kind of frustrating when you get adds and spam email because of things that your teenager and a bunch of their friends converse with Google Home about. It has not been used for over 2 years and I still get some trailing adds from things communicated to that device. Unsettling. I know my phone is doing pretty much the same thing, but at least I am the only one that uses it so I can control the type of content that is searched for.

-----Original Message-----
From: Steve Lee <stevelee@w3.org> 
Sent: Wednesday, February 3, 2021 4:35 AM
To: public-wai-cc@w3.org
Subject: Re: Clarifying speech recognition vs. voice recognition in Style Guide

 > "speech recognition" is for speech-to-text

Though Alexa etc and voice menus use it to determine user intent, possibly via an internal STT step.

Another possible wrinkle is that Alexia can be trained to respond to various people - I do not know if it recognises them though or just adjusts to all of them.

Steve

On 02/02/2021 14:43, Shawn Henry wrote:
> Hi folks,
> 
> Here is a draft update to the WAI Style Guide[1]:
> 
> "speech recognition" is for speech-to-text (SST), and usually what 
> we're talking about for accessibility.
> "voice recognition" is different; it's about identifying the speaker, 
> not what they're saying.
> 
> (For background, search the Web for "speech recognition voice 
> recognition difference")
> 
> Please let me know if you disagree or have edit suggestions.
> 
> Thanks!
> ~Shawn
> 
> [1] currently at the end of
> https://urldefense.proofpoint.com/v2/url?u=https-3A__www.w3.org_WAI_EO

> _wiki_Style-23other-5Fwords-5Fand-5Fphrases&d=DwICaQ&c=0YLnzTkWOdJlub_
> y7qAx8Q&r=v-L6X-ScaY5UKb-F-_zcuXdbPw2UYK_gaTG8R5d9h7U&m=tTtRkat9YrnNNl
> hg6dtEbeqLNp51_LseZ1pdFBaqeZ4&s=mFtxbR59MlcEk949G95EV1ayjPBmglYNbKCPaw
> uLczU&e=
> 

Received on Wednesday, 3 February 2021 13:59:28 UTC

This archive was generated by hypermail 2.4.0 : Wednesday, 3 February 2021 13:59:29 UTC