Presentation Title

Author

Why can’t Siri pronounce my name?

Pros and cons of the three modes of communication: speech, sign, writing
Writing is the representation of language with the use of visual signs
- So, not pictograms
- Iconic vs. symbolic
- Phonological or morphological
Writing is learned
- But certain aspects of computer-mediated communication (CMC) seem like they are acquired
- Emojis as extralinguistic material (like facial expressions and body language)
What makes the digital environment different from face-to-face?

Computational linguistics is the area of linguistics that computer scientists like…
- Linguistics + computers = you can rule the world!
- If you are looking for the most lucrative field, this is it!
Computational linguistics is about:
- Speech synthesis
- Automatic (computer) recognition of speech
- Machine translation
- Corpus construction

Where can you see the intersection of linguistics and computer science in your everyday lives?

One thing that has become ever-present in our lives is the digital world
- You can speak your text message to your phone – this is called speech-to-text
- You can have your text messages read aloud by your phone – this is called text-to-speech
Speech technologies are increasingly integrated into our lives!
- Everyone is obsessed with AI
- There are refrigerators, cars, etc. with built-in speech technologies
- Your phone might have a virtual assistant, like Siri

Your devices rely on speech recognition technologies , which means that they have the ability to recognize and translate human spoken language into text that can be understood by computers
- This allows you to use commands like “call Mom”
- Has a wide array of applications
They also rely on speech synthesis , which is the artificial production of human speech by a computer
- They sound so real now
- Here’s how it used to sound…

True or false: In 1962, Bell Labs programmed a computer to speak and sing! True

Voice search

Compositing text

Turning on the lights

Customer service applications

Biometrics

In-car features

Duolingo

Speech recognition is complex
- Background noise can get in the way
- Linguistic variation can be tricky to capture
I said they were getting better – how do we know?
- Accuracy – how often does the system get it wrong?
- Speed – how quickly does it convert from speech to text?

Which one is better for accuracy – speaker-dependent or speaker-independent? Dependent

Has Siri had trouble understanding what you’ve said?
Some of the accuracy issues may have been part of the problem, but often, linguistic variation in accent, word choice, etc. can be the problem
Research has shown that
- African Americans experience error rates up to two times higher than other speakers of American English
  - Word identification is less accurate, but
  - Word prediction is sometimes better compared to majority dialects
- Linguists use “force-aligners” to automatically transcribe speech, and one study found that error rates on accuracy were higher for Southern varieties compared to Northern ones

Proportion of correctly recognized words by dialect area, color coded by country
- Used YouTube auto-captions on Accent Tag videos (powered by Google)
- Works best on people from California (which just happens to be where Google is headquartered)
- Works worst on Scottish English ::: {.notes}

https://makingnoiseandhearingthings.com/2016/07/11/which-accents-does-automatic-speech-recognition-work-best-for/

:::

Two (Three?) words:

Self-driving cars

But also – all the things we can do with speech recognition and speech synthesis will only grow

We will all want/need to be able to operate in that future!

Come next week for more fun with computers!
- “How do computers help with language?”
- “Are chatbots real?”
Quiz 9 in recitation on Friday
Observation notes #6 (last one!) due Friday, April 12 at 11:59pm
- No late work, no AI
- Drop lowest – if you are happy, you can skip it!