Min menu

Pages

Speechmatics outperforms Google in voice recognition

 A British speech recognition startup said its voice recognition technology was better than big tech companies like Google and Amazon at understanding the voices of black people.

A British speech recognition startup said its voice recognition technology was better than big tech companies like Google and Amazon at understanding the voices of black people.   Speechmatics said its system has an overall accuracy rate of 83 percent for African American votes.  This is higher than Microsoft (73 percent), Amazon (69 percent), Google (69 percent), and Apple (55 percent), according to research published by Stanford University in 2020.  The Stanford University research compared results from major tech companies on how accurately speech recognition software understands African Americans.  On top of that, Amazon, Google, Microsoft and Apple systems made nearly twice as many errors when interpreting words spoken by African Americans than whites, according to researchers at Stanford University.  Speechmatics says its system misrecognizes words from black voices 17% of the time, compared to 31% for Google and Amazon.  “It is critical to study and improve equity in speech-to-text systems because of the potential for disproportionate harm to individuals through downstream sectors ranging from health care to criminal justice,” said Alison Koenicki, lead author of the Stanford study.  Voice recognition technology has become an integral part of everyday life, thanks to the proliferation of virtual assistants via smart devices such as phones and speakers.  Apple pioneered the use of voice-activated software on mobile devices through its digital assistant, Siri.  While Amazon was one of the first companies to bring speech recognition into the home with its Echo speakers and Alexa assistant.  Bias in voice recognition technology Researchers are becoming increasingly concerned about bias in the algorithms that support these speech recognition services.  Experts say many voice-recognition programs are trained on limited sets of data, making them less effective.  It is related to the quality of the data in the training sets. And there has been racial bias, gender bias, and regional dialect bias in speech recognition technology for a long time. And this technology doesn't work the same way for everyone yet.  Speechmatics says it trained its AI with unclassified data from social media and podcasts. This is to help her learn different aspects of speech including dialect and language.  On top of that, the company said its technology is trained for 1.1 million hours of audio. Speechmatics described the development as a breakthrough.  She hopes that other tech companies will become more transparent about efforts to reduce bias in AI.  As a result, tech giants have ramped up their investments in speech recognition recently. Microsoft agreed to acquire software company Nuance Communications for $16 billion in April.


Speechmatics said its system has an overall accuracy rate of 83 percent for African American votes.


This is higher than Microsoft (73 percent), Amazon (69 percent), Google (69 percent), and Apple (55 percent), according to research published by Stanford University in 2020.


The Stanford University research compared results from major tech companies on how accurately speech recognition software understands African Americans.


On top of that, Amazon, Google, Microsoft and Apple systems made nearly twice as many errors when interpreting words spoken by African Americans than whites, according to researchers at Stanford University.


Speechmatics says its system misrecognizes words from black voices 17% of the time, compared to 31% for Google and Amazon.


“It is critical to study and improve equity in speech-to-text systems because of the potential for disproportionate harm to individuals through downstream sectors ranging from health care to criminal justice,” said Alison Koenicki, lead author of the Stanford study.


Voice recognition technology has become an integral part of everyday life, thanks to the proliferation of virtual assistants via smart devices such as phones and speakers.


Apple pioneered the use of voice-activated software on mobile devices through its digital assistant, Siri.


While Amazon was one of the first companies to bring speech recognition into the home with its Echo speakers and Alexa assistant.


Bias in voice recognition technology

Researchers are becoming increasingly concerned about bias in the algorithms that support these speech recognition services.


Experts say many voice-recognition programs are trained on limited sets of data, making them less effective.


It is related to the quality of the data in the training sets. And there has been racial bias, gender bias, and regional dialect bias in speech recognition technology for a long time. And this technology doesn't work the same way for everyone yet.


Speechmatics says it trained its AI with unclassified data from social media and podcasts. This is to help her learn different aspects of speech including dialect and language.


On top of that, the company said its technology is trained for 1.1 million hours of audio. Speechmatics described the development as a breakthrough.


She hopes that other tech companies will become more transparent about efforts to reduce bias in AI.


As a result, tech giants have ramped up their investments in speech recognition recently. Microsoft agreed to acquire software company Nuance Communications for $16 billion in April.

reaction:

Comments