Week in voice AI #7
Apple is into whispering speech
I am Ivan Mehta, a consumer tech reporter at TechCrunch. For more than a year, I have covered different aspects of voice AI. This newsletter is an experimental attempt at covering what is happening in this industry, which is growing at a rapid pace.
Note: This edition was supposed to go out on Monday (February 2), but being at a conference didn’t allow me to finish this edition. While a lot has happened already, including ElevenLabs’ mega funding round, I’ll write more about that next week.
Google pays $68 million in a lawsuit alleging the company was listening to consumer conversations
The biggest news of the week was no doubt Google settling a lawsuit against it that alleged that the company’s Google Assistant listened to users. While the company is paying $68 million, it is not accepting any wrongdoing.
The plaintiffs allege that Google “listened” to consumers without their consent and disclosed those communications to third parties.
“Plaintiffs in the lawsuit allege that Google Assistant can activate and record communications even when a user does not intentionally trigger Google Assistant with a hot word, like “Okay Google,” or manually activate Google Assistant on their device. Plaintiffs refer to such instances as “False Accepts” and allege that Google wrongfully collects, uses, and discloses user audio recordings from False Accepts to improve the speech recognition abilities of Google Assistant,” the case document reads.
The complainants aren’t mincing any words and are directly pointing fingers at Google for allegedly using this data for targeted advertising.
To be sure, Google’s violations are deliberate and calculated to lead to increased revenues for Google. Google profits on this invasion of privacy by using the content of conversations Google admits it obtains without consent or authorization to improve the functionality of Google Assistant and to target personalized advertising to the very consumers whose privacy has been invaded.
2 Google can monetize its platform by using its Google Assistant ecosystem to drive revenue to complementary lines of business such as e-commerce, search, or entertainment. Improving the functionality of Google Assistant is necessary to allow Google to continue to compete with the increasingly crowded virtual assistant market and to buttress the price of the Google Manufactured Devices. In short, Google intentionally, willfully, and knowingly violated Plaintiffs’ and consumers’ privacy rights, including within the sanctity of Plaintiffs’ and consumers’ own homes where they have the greatest expectation of privacy
“Devices are listening to us” is not a new theory in any shape or form. But cases like these, even if there is no admission of guilt, seed more doubt in the minds of customers and privacy watchdogs.
More companies want to make voice a prominent interface for interaction, and in tryst for that, we now have more devices with a mic around us. This means our conversations are more prone to privacy violations. The companies will need to bolster their security provisions and ability to fight privacy claims.
Apple acquires Q.ai
Apple acquired Israel-based voice AI company Q.ai for a reported price of $2 billion. While the startup was largely secretive about the work it did, it worked on understanding whispering speech. Plus, it also worked on reducing noise from captured audio.
While Apple is working on a new Siri, the audio experience is not always great. The assistant often misunderstands what users are saying. With this acquisition, Apple will likely invest in making Siri understand you better. Plus, it might also work to improve interaction through AirPods and Vision Pro.
Report highlights the importance of unique and emotional voices
A new report from voice marketplace Voices said that 29% of the companies it surveyed had deployed voice AI. Among 200 decision makers, 77% of them said that exclusive voice rights matter to them. 61% ofthe consumers the company interviewed also said that if the voices are exclusive, they are more memorable.
More than two-thirds of companies insisted on using voice actors to create the base for the AI voices rather than using a voice generator. But this might come down to preference and cost, as voice model companies are trying to make generations sound more “human.”
Quick bytes
Deezer is selling its music detection tech as the company said daily delivery of AI tracks uploaded on the platform was 39% in January. The company also said that up to 85% streams on AI-generated tracks are fraudulent, and are being tried to game the system to monetize these tracks.
Music publisher Wixen sued Meta and said that the platform is trying to replace human creators with AI-generated music.
A group of music publishers, including Universal Music Group, sued Anthropic, alleging that the AI company used their work to train Claude without permission.
Productivity company Genspark releases a new dictation app called Speakly, which also works on Genspark for agentic commands.
Sudha Murty warned users about financial scams that use her voice and face.
Voice coaching platform BoldVoice, which helps non-native English speakers, raised $21 million in Series A led by Matrix.
Signals & Experiments
I was at Web Summit Qatar with Comulytic’s (the name could have been better) hardware note taker, and I also tested out Plaud’s NotePin S. Both are pretty handy devices, but there are some quirks that could be improved.
The NotePin S has a hardware button, but the light indicator on the device is very small. Every time I turned it on, I had to check multiple times that it was recording. I didn’t miss any recordings, but I wish there were a better indicator.
Comulytic’s battery was a tad disappointing. The company claims that the device has 45 hours continous recording in it with 107 days of standby time. But I was nowhere near 45 hours of recording or 107 since I unboxed the device.
I expected to see more folks using hardware notetakers at a tech conference. But the only person I saw using one was my co-panelist and WebFlow CEO, Linda Tong, who was using Pocket.
Thank you for tuning in. Keep listening.
Email: voiceaiweek@gmail.com or im@ivanmehta.com
.