Speech Technology Predictions for 2019
We asked our regular contributors and columnists what speech technology trends and stories will dominate the headlines in 2019. Which of these predictions do you think will come true in the year to come, and which do you think we’ll still be waiting for in 2020?
James A. Larson, Vice President of Larson Technical Services and Chair of the SpeechTEK Conference says:
“Voice morphing, the substitution of one voice by another, will be in the news:
- Voice morphing software will be available for mobile devices
- Bullies will use voice morphing to confuse and bully their victims
- False news reports will use voice morphing to simulate the voices of authorities
- Voice databases will be hacked so fraudsters will have access to voices for training voice morphing software.
- Regularity agencies will be slow to establish and enforce voice piracy laws
- Users will hang up when they hear ‘This conversation will be recorded for quality assurance’ to avoid their voices being captured and placed in voice databases”
Deborah Dahl, Principal at Conversational Technologies, says:
“I think the hottest trend for 2019 will be that more and more applications will be capable of identifying and reacting to users' emotions. Identifying emotions from speech and language is technically possible now, but more interestingly, it raises some important design questions. What should the system say to an angry or upset caller? Should it try to calm down the caller, or will that just make things worse? Application designers will also have to take the confidence of the emotion recognition result into account -- it would be bad if the system mistook a neutral emotion for anger and began apologizing!”
Kevin Brown, Managing Director at Vox Peritus says:
“Significant increases in the use of artificial neural networks (ANN) within enterprises will happen in 2019.
User acceptance of speech recognition has increased dramatically through artificial neural networks provided by Google, Microsoft, Apple, Amazon among others. The marked dichotomy between consumer speech recognition and currently much worse enterprise speech recognition capabilities demands a rapid change in approach.
For 2019, ANN will be added to create hybrids with traditional speech recognition capabilities until third party ANN providers choose their respect niche industries. The result will be opportunities for new players to enter the enterprise speech recognition marketplace.”
Dr. Moshe Yudkowsky, President of Disaggregate, says:
“The European Union now mandates that all uploads of digital content be scanned for alleged copyright violations--in other words, pre-emptive censorship. This includes any content, even code uploaded to open-source software projects. In 2019, expect this mandate to expand to speech recognition (ASR) in order to censor disfavored or unpopular opinions, those disparaged as ‘hate speech.’ The EU or the UK may mandate it. In the US, social media companies may turn to ASR to supplement their current army of human censors.”
The technology is magical, but can be misused
Planned Parenthood introduced Roo, a chatbot that answers the sometimes embarrassing questions of teens and young adults who might otherwise not have access to science-based sexual education.
Speech dictation has been found to be three times faster than touch screen typing on mobile devices and two to three times faster than typing on a full keyboard. And the cost of obtaining an automated transcript is usually quite low. Not to mention the quick turnaround time. There are a number of ways speech technology can make processes in our daily lives and work environments more efficient.