5 ms·
I tried and failed to launch a startup in this space. The privacy concerns are huge. Most of the companies in this space (Fireflies, Otter, Vowel, ~Chorus, ~Gon
by axg11 5y ago
I tried and failed to launch a startup in this space. The privacy concerns are huge. Most of the companies in this space (Fireflies, Otter, Vowel, ~Chorus, ~Gong) use external APIs for speech-to-text. That means unencrypted audio and plain text is traversing a 3rd party server. Most people don't care, but when there is eventually a security breach (only a matter of time) there will be a lot of surprised users and customers.
With that said, the core idea is great. These companies are starting with transcription but it's not difficult to envisage a future where (a) supplemental information is presented depending on conversation topics, and (b) actions are taken on your behalf. Example: no more need to manually send a calendar invite, just mention it in the conversation and the "AI assistant" will schedule for you.
- mikepurvis 5y agoI had a family member who briefly dabbled in online transcription and made maybe a few hundred bucks doing it. Obviously they were required to sign an NDA and so on, but nonetheless, I was shocked at the recordings that they were sent— for example, high level executive meetings where financial details and M&A strategy were being discussed.
- jfengel 5y agoI'm shocked that you got to hear those recordings that were supposedly under NDA. Well, not that shocked. I'm sure your uncle was generally being diligent. But really, you shouldn't have been able to find out what he was working on.
- mikepurvis 5y agoI wasn't hearing them directly, but I heard about them because I was asked about some acronyms that were unfamiliar to the person doing the work. Definitely a grey area, but the point was more that literally anyone off the street can sign up to gig for one of these services and receive access to a bunch of potentially pretty sensitive audio.
- Spooky23 5y agoEven stranger, in many legal scenarios, verbatim reporters actually own the transcripts and license/sell copies to courts. Some of those transcripts are sealed or otherwise private. If you have good relationships in the courthouse you can luck out with complex cases that result in appeals.
- skoocda 5y agoI also tried and failed here. We ran our own speech engine with a custom model- but it's extremely expensive as a cloud service, and incredibly tough to reach acceptably high accuracy in different environments. Adding NLP on top of error-prone transcripts will multiply the error rate and lead to all sorts of weird actions. I really think on-device models like we see in Android's Live Caption tool are a major privacy boon, and they're starting to reach an acceptable level of performance in Google's case. The main pathway to better performance is loading ever-more-massive models into memory, which isn't feasible for mobile devices but could be done on people's laptops in a meeting.
- toyg 5y ago"Privacy", "on-device", and "Google", all in the same sentence...? Sounds unrealistic.
- deadmutex 5y agoPichai: "Yes, we use data to make products more helpful for everyone. But we also protect your information." https://www.nytimes.com/2019/05/07/opinion/google-sundar-pichai-privacy.html https://www.nytimes.com/2019/05/07/opinion/google-sundar-pic...
- ai_ia 5y agoI believe you ought to check out Federated Machine Learning techniques.[1] [1]: https://ai.googleblog.com/2017/04/federated-learning-collaborative.html https://ai.googleblog.com/2017/04/federated-learning-collabo...
- dillondoyle 5y agoWould it not be possible to 'ship' a model and have it run on the users device?
- junon 5y agoThis is why I stopped using Otter, personally. I was more worried about IP theft than anything personally sensitive but yeah. I don't use speech to text solutions anymore unless they can somehow run on my own machines.