Speech recognition systems leave out a large demographic of English speakers because they can only recognise accents they’ve been trained to understand. Getty
Speech recognition systems leave out a large demographic of English speakers because they can only recognise accents they’ve been trained to understand. Getty
Speech recognition systems leave out a large demographic of English speakers because they can only recognise accents they’ve been trained to understand. Getty
Speech recognition systems leave out a large demographic of English speakers because they can only recognise accents they’ve been trained to understand. Getty

Accents and AI: how speech recognition software could lead to new forms of discrimination


  • English
  • Arabic

Anyone who has used a voice assistant such as Apple's Siri or Amazon's Alexa will have occasionally struggled to make themselves understood. Perhaps the device plays the wrong music, or puts unusual items on a shopping list, or emits a plaintive “didn't quite catch that”. But for people who speak with an accent, these devices can be unusable.

The inability of speech recognition systems to understand accents found in Scotland, Turkey, the southern states of the US or any number of other places is widely documented on social media, and yet the problem persists. With uses of the technology now spreading beyond the domestic, researchers and academics are warning that biased systems could lead to new forms of discrimination, purely because of someone’s accent.

“It's one of the questions that you don't see big tech responding to,” says Halcyon Lawrence a professor of technical communication at Towson University in Maryland who is from Trinidad and Tobago. “There's never a statement put out. There's never a plan that's articulated. And that's because it's not a problem for big tech. But it’s a problem for me, and large groups of people like me.”

Speech recognition systems can only recognise accents they’ve been trained to understand. To learn how to interpret the accent of someone from Trinidad, Eswatini or the UAE, a system needs voice data, along with an accurate transcription of that data, which inevitably has to be done by a human being. It’s a painstaking and expensive process to demonstrate to a machine what a particular word sounds like when it’s spoken by a particular community, and perhaps inevitably, existing data is heavily skewed towards English as typically spoken by white, highly educated Americans.

If you plot new accent releases on a map, you can’t help but notice that the Global South is not a consideration, despite the numbers of English speakers there
Halcyon Lawrence,
a professor of Technical Communication at Towson University in Maryland

A study called Racial Disparities in Automated Speech Recognition, published last year by researchers at Stanford University, illustrates the stark nature of the problem. It analysed systems developed by Amazon, Apple, Google, IBM and Microsoft, and found that in every case the error rates for black speakers were nearly double that of white people. In addition, it found that the errors were not caused by grammar, but by “phonological, phonetic, or prosodic characteristics”; in other words, accent.

Allison Koenecke, who led the study, believes that a two-fold improvement in the system is needed. “It needs resources to ethically collect data and ensure that the people working on these products are also diverse,” she says. “While tech companies may have the funds, they may not have known that they needed to prioritise this issue before external researchers shone a light on it.”

Lawrence, however, believes that the failings are no accident.

“What, for me, shows big tech's intention is when they decide to release a new accent to the market and where that is targeted,” she says. “If you plot it on a map, you can’t help but notice that the Global South is not a consideration, despite the numbers of English speakers there. So you begin to see that this is an economic decision.”

It’s not only accented English that scupper speech recognition systems. Arabic poses a particular challenge – not simply because of the many sub-dialects, but inherent difficulties such as the lack of capital letters, recognising proper nouns and predicting a word’s vowels based on context. Substantial resources are being thrown at this problem, but the current situation is the same as with English: large communities technologically disenfranchised.

Why is this of particular concern? Beyond the world of smart speakers lies a much bigger picture. “There are many higher-stakes applications with much worse consequences if the underlying technologies are biased,” says Koenecke. “One example is court transcriptions, where court reporters are starting to use automatic speech recognition technologies. If they aren't accurate at transcribing cases, you have obvious repercussions.”

Lawrence is particularly concerned about the way people drop their accent in order to be understood, rather than the technology working harder to understand them. “Accent bias is already practiced in our community,” she says. “There's an expectation that we adapt our accent, and that's what gets replicated in the device. It would not be an acceptable demand on somebody to change the colour of their skin, so why is it acceptable to demand we change our accents?”

Money, as ever, lies at the root of the problem. Lawrence believes strongly that the market can offer no solution, and that big tech has to be urged to look beyond its profit margin. “It’s one of the reasons why I believe that we’re going to see more and more smaller independent developers do this kind of work,” she says.

One of those developers, a British company called Speechmatics, is at the forefront, using what it calls “self-supervised learning” to introduce its speech recognition systems to a new world of voices.

If you have the right kind of diversity of data, it will learn to generalise across voices, latch on quickly and understand what's going on
Will Williams,
vice president of Machine Learning

“We're training on over a million hours of unlabelled audio, and constructing systems that can learn interesting things, autonomously run,” says Will Williams, vice president of machine learning at Speechmatics.

The crucial point: this is voice data that hasn’t been transcribed. “If you have the right kind of diversity of data, it will learn to generalise across voices, latch on quickly and understand what's going on.” Using datasets from the Stanford study, Speechmatics has already reported a 45 per cent reduction in errors when using its system.

An organisation called ML Commons, which has Google and Microsoft as two of its more than 50 founding members, is now looking for new ways to create speech recognition systems that are accent-agnostic.

It’s a long road ahead, but Koenecke is optimistic. “Hopefully, as different speech-to-text companies decide to invest in more diverse data and more diverse teams of employees such as engineers and product managers, we will see something that reflects more closely what we see in real life.”

Sole survivors
  • Cecelia Crocker was on board Northwest Airlines Flight 255 in 1987 when it crashed in Detroit, killing 154 people, including her parents and brother. The plane had hit a light pole on take off
  • George Lamson Jr, from Minnesota, was on a Galaxy Airlines flight that crashed in Reno in 1985, killing 68 people. His entire seat was launched out of the plane
  • Bahia Bakari, then 12, survived when a Yemenia Airways flight crashed near the Comoros in 2009, killing 152. She was found clinging to wreckage after floating in the ocean for 13 hours.
  • Jim Polehinke was the co-pilot and sole survivor of a 2006 Comair flight that crashed in Lexington, Kentucky, killing 49.
Who are the Soroptimists?

The first Soroptimists club was founded in Oakland, California in 1921. The name comes from the Latin word soror which means sister, combined with optima, meaning the best.

The organisation said its name is best interpreted as ‘the best for women’.

Since then the group has grown exponentially around the world and is officially affiliated with the United Nations. The organisation also counts Queen Mathilde of Belgium among its ranks.

The burning issue

The internal combustion engine is facing a watershed moment – major manufacturer Volvo is to stop producing petroleum-powered vehicles by 2021 and countries in Europe, including the UK, have vowed to ban their sale before 2040. The National takes a look at the story of one of the most successful technologies of the last 100 years and how it has impacted life in the UAE. 

Read part four: an affection for classic cars lives on

Read part three: the age of the electric vehicle begins

Read part one: how cars came to the UAE

 

THE BIO

Favourite place to go to in the UAE: The desert sand dunes, just after some rain

Who inspires you: Anybody with new and smart ideas, challenging questions, an open mind and a positive attitude

Where would you like to retire: Most probably in my home country, Hungary, but with frequent returns to the UAE

Favorite book: A book by Transilvanian author, Albert Wass, entitled ‘Sword and Reap’ (Kard es Kasza) - not really known internationally

Favourite subjects in school: Mathematics and science

Company profile

Name: Infinite8

Based: Dubai

Launch year: 2017

Number of employees: 90

Sector: Online gaming industry

Funding: $1.2m from a UAE angel investor

MATCH INFO

Uefa Champions League semi-final, first leg

Barcelona v Liverpool, Wednesday, 11pm (UAE).

Second leg

Liverpool v Barcelona, Tuesday, May 7, 11pm

Games on BeIN Sports

Planes grounded by coronavirus

British Airways: Cancels all direct flights to and from mainland China 

Hong Kong-based Cathay Pacific: Cutting capacity to/from mainland China by 50 per cent from Jan. 30

Chicago-based United Airlines: Reducing flights to Beijing, Shanghai, and Hong Kong

Ai Seoul:  Suspended all flights to China

Finnair: Suspending flights to Nanjing and Beijing Daxing until the end of March

Indonesia's Lion Air: Suspending all flights to China from February

South Korea's Asiana Airlines,  Jeju Air  and Jin Air: Suspend all flights

The biog

Name: Salvador Toriano Jr

Age: 59

From: Laguna, The Philippines

Favourite dish: Seabass or Fish and Chips

Hobbies: When he’s not in the restaurant, he still likes to cook, along with walking and meeting up with friends.

Tips for newlyweds to better manage finances

All couples are unique and have to create a financial blueprint that is most suitable for their relationship, says Vijay Valecha, chief investment officer at Century Financial. He offers his top five tips for couples to better manage their finances.

Discuss your assets and debts: When married, it’s important to understand each other’s personal financial situation. It’s necessary to know upfront what each party brings to the table, as debts and assets affect spending habits and joint loan qualifications. Discussing all aspects of their finances as a couple prevents anyone from being blindsided later.

Decide on the financial/saving goals: Spouses should independently list their top goals and share their lists with one another to shape a joint plan. Writing down clear goals will help them determine how much to save each month, how much to put aside for short-term goals, and how they will reach their long-term financial goals.

Set a budget: A budget can keep the couple be mindful of their income and expenses. With a monthly budget, couples will know exactly how much they can spend in a category each month, how much they have to work with and what spending areas need to be evaluated.

Decide who manages what: When it comes to handling finances, it’s a good idea to decide who manages what. For example, one person might take on the day-to-day bills, while the other tackles long-term investments and retirement plans.

Money date nights: Talking about money should be a healthy, ongoing conversation and couples should not wait for something to go wrong. They should set time aside every month to talk about future financial decisions and see the progress they’ve made together towards accomplishing their goals.

F1 The Movie

Starring: Brad Pitt, Damson Idris, Kerry Condon, Javier Bardem

Director: Joseph Kosinski

Rating: 4/5

The Vines - In Miracle Land
Two stars

Benefits of first-time home buyers' scheme
  • Priority access to new homes from participating developers
  • Discounts on sales price of off-plan units
  • Flexible payment plans from developers
  • Mortgages with better interest rates, faster approval times and reduced fees
  • DLD registration fee can be paid through banks or credit cards at zero interest rates

England's all-time record goalscorers:
Wayne Rooney 53
Bobby Charlton 49
Gary Lineker 48
Jimmy Greaves 44
Michael Owen 40
Tom Finney 30
Nat Lofthouse 30
Alan Shearer 30
Viv Woodward 29
Frank Lampard 29

Dubai Bling season three

Cast: Loujain Adada, Zeina Khoury, Farhana Bodi, Ebraheem Al Samadi, Mona Kattan, and couples Safa & Fahad Siddiqui and DJ Bliss & Danya Mohammed 

Rating: 1/5

UAE currency: the story behind the money in your pockets
RIVER%20SPIRIT
%3Cp%3E%3Cstrong%3EAuthor%3A%20%3C%2Fstrong%3ELeila%20Aboulela%C2%A0%3C%2Fp%3E%0A%3Cp%3E%3Cstrong%3EPublisher%3A%3C%2Fstrong%3E%20Saqi%20Books%3C%2Fp%3E%0A%3Cp%3E%3Cstrong%3EPages%3A%3C%2Fstrong%3E%20320%3C%2Fp%3E%0A%3Cp%3E%3Cstrong%3EAvailable%3A%3C%2Fstrong%3E%20Now%3C%2Fp%3E%0A
Updated: November 07, 2021, 2:54 PM`