Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple Training Siri to Better Understand People With Atypical Speech
- Thread starter MacRumors
- Start date
- Sort by reaction score
Understanding people with atypical speech patterns?!? My God, Siri rarely understands ME — and I've got a very clear and understandable voice.
Apple is researching how to improve Siri to better understand people who talk with a stutter, according to new details shared by The Wall Street Journal in a piece on how companies train voice assistants to handle atypical speech.
Apple has built a bank of 28,000 audio clips from podcasts featuring people who stutter, which could be used to train Siri. The data that Apple has collected will improve voice recognition systems for atypical speech patterns, according to an Apple spokesperson.
Along with improving how Siri understands people with atypical speech patterns, Apple has also implemented a Hold to Talk feature for Siri that allows users to control how long they want Siri to listen for. This prevents Siri from interrupting users with a stutter before they're finished speaking.
Siri can also be used without voice all together, through a Type to Siri feature that was first introduced in iOS 11.
Apple plans to outline its work to improve Siri in a research paper set to be published this week, which will provide more details on the company's efforts.
Google and Amazon are also working to train Google Assistant and Alexa to better understand all users, including those who have trouble using their voices. Google is collecting atypical speech data, and Amazon in December launched the Alexa Fund to let people who have speech impairments train an algorithm to recognize their unique vocal patterns.
Article Link: Apple Training Siri to Better Understand People With Atypical Speech
Me: ‘Hey Siri, Play BBC radio 2’
Siri: ‘Sorry, I can’t play that radio station ‘
Me: ‘Hey Siri, Play radio 2 by the BBC’
Siri: ‘Here’s BBC radio 2, provided by tune in’
BBC radio 2 plays. 😱😳
Siri: ‘Sorry, I can’t play that radio station ‘
Me: ‘Hey Siri, Play radio 2 by the BBC’
Siri: ‘Here’s BBC radio 2, provided by tune in’
BBC radio 2 plays. 😱😳
Well, you may not consider it atypical, but stuttering is not how people talk, typically (atypical meaning not-typical). Additionally, stuttering is an example of atypical speech, but it’s not the ONLY kind of atypical speech. So, they can either list out all the different atypical speech patterns they attempting to correct for OOOOOR they can just say “atypical speech”.
Hope this helps with Siri requiring you to repeat some words that Siri gets easily confused with.
Maybe (probably) I'm in the minority, but my Siri rarely has difficulty understanding me. I'm not discounting all those who's doesn't...I'm sure that is frustrating. Im all for any improvements to software.
How about they train Siri to understand accents. Better yet, make it able to learn accents.
It's supremely frustrating how persistently Siri refuses to understand me even if I repeat the same phrase 10 times in a row because I have an accent. Touch Id can learn my fingerprints, Face ID can learn my face but almost 10 years after its release Siri can't learn how I pronounce anything.
I gave up on Siri long agofor all but the simplest tasks. Whenever I try anything else it's hit or miss (mostly miss).
It's supremely frustrating how persistently Siri refuses to understand me even if I repeat the same phrase 10 times in a row because I have an accent. Touch Id can learn my fingerprints, Face ID can learn my face but almost 10 years after its release Siri can't learn how I pronounce anything.
I gave up on Siri long agofor all but the simplest tasks. Whenever I try anything else it's hit or miss (mostly miss).
Just what I think, but I think part of it is because Apple does not do the data collecting and care for privacy more. Google's Assisstant and Amazon Alexa probably has the mic on 24/7 for everyone out there training their AI on all the youtube videos, voice recordings, audio calls ...etc . Maybe I am wrong, but just my theory on it.
I hope Siri gets better. I have a speech impairment. And it gets frustrating I have to repeat myself at times when im trying to get Siri to type something for me when I do the audio to text thing to send a message from my apple watch. I lost count I will even say Hey Siri to many times even then she don't get me. Blast them r's...
28.000 sounds a lot, but in reality it's not much. "The word gap" is a real thing. By the age of 5 children who get stories and are engaged in conversation are exposed to 4 to 30 million more words than children who are not.
How about Apple direct some of those resources to making Siri understand people with "typical" speech. Apple's fetish for exceedingly niche use cases/things/etc. is compromising their core products.
I mean, good luck. But to my knowledge my speech is perfectly typical, and I can speak english natively without an accent, and Siri still doesn't understand me...
Thank you for posting that - I don't see many videos of people who stutter, and I stutter sometimes in a very similar fashion. I didn't really know anyone else who stuttered until about a year ago when a new colleague joined the company who has a very, what I'd call "traditional" stutter that you see in some of the older TV shows. His stutter affects him much prominently than mine does, which I can usually work through by pausing or 'pushing through'. Made me feel a little less unique/alone.
As a user of Siri in a foreign language (italian), I always find difficult to ask for an english song or an album title. Simply, I must repeat the title pronouncing it as it was italian. Which is at the same time ridicolous and unreliable. Made it simple, Alexa is years ahead in this and other features that imply the combination of different languages (or in any case the use of a non-standard language word).
