Why would for TextHear only the Android app be free?
It’s hard to “get” the use-cases from that article. I’m really confused how they differ (or how they are similar). Like, if you are transcribing a phone call, then how do you look at your phone? You have to use speaker? Or if it’s a landline, then is the phone basically used as a display for the HOH person? Or what about face to face conversations? Do any of these do that like Live Transcribe does? Hopefully someone breaks these into a table and compares them at some point. Cool that they exist though. Before Live Transcribe, the only assistance I’ve used is CaptionCall, and asking people to repeat themselves.
Why have only four apps been examined? On Google Play there are perhaps 100 apps, of which at least 30 claim to deliver what I want - simply accurate transcription of what is being said around me. Most of the poor reviews comment upon the absence of capability to retain the text as a text file. Useful, I admit, but not essential. Many of the offerings can translate foreign languages into English, or vice versa - I do not need this. What I do need to know is - which of the 30 “possibles” is the most accurate?
A new app, Microsoft Translater, needs to be added to this list. It was developed to translate from language to language, but has now been set up for speech recognition. It is most similar to AVA, as it is designed for person to person Speech, but can also be used like Google Live Transcript to recognize voices around the phone and Transcribe them. It is free and available on all phones, Apple or Android, and also computers, and it’s free. But it’s not as fast or accurate as Live Transcribe.
A version of Microsoft Translator is now included in PowerPoint so that the speaker is captioned. It’s a feature you just turn on. The captions can be positioned at top or bottom of slides. Google Live Transcribe is also included in Google Slides to provide captioning, but the captions are only at the bottom of the slides.
Given my level of success from using speech-to-text with texting, this talk of using it in conversation brings up memories of the Monty Python English-Hungarian phrasebook.
My iPhone doesn’t understand my Texas accent and back when I thought I could use speech-to-text it was frequently hilarious
Google Live Transcribe is using machine learning by inputting for various accents. So you may want to try it again
Thanks! I will try it again.
Another new speech to text I’ve tried is Otter. It’s fast and accurate like Google Live Transcribe, but works on any device. It also produces a Transcript and can learn to identify the speakers on the Transcript.
Thank you very much!
Having tried both Ava and Otter, I have for the time being given up in my efforts to use speech-to-text as a means of finding out what people are saying in small meetings of 6 to 12 people sitting around a table. One factor which no machine can cope with is connected to the fact that spoken English is quite different to written English both in the words used and in the way they are arranged. In 1 to 1 situations where you are facing the person you are listening to, using a good hearing aid, it is quite easy for your brain to edit out the unwanted and useless bits - things like “sort of” and “yeah” and phrases which are instantly corrected, and to edit in words that are mispronounced and the ends of phrases that are deliberately put across at lower volume. No electronic machine can do this - all it can do is examine each word as it is uttered and try to find a written word that appears to fit it and print it out. Sometimes it looks like rubbish because it is rubbish; sometimes it is difficult to know what it means because it has been put together in a disjointed fashion. Sometimes the machine seems to think that its own predictive texting will be better than the jumbled phrasiology that has been spoken. This is usually a bad mistake and the result is worse than ever. If some apps are better than others, I would be happy to try them out, but I am not optimistic. Life is too short to go through all the 100+ apps in google play, only to find that most, if not all of them cannot cope with the sort of speech that people use in small meetings.
I haven’t tested it yet, but sounds promising: New AI-Powered Hearing App ‘HeardThat’ to Rise Above the Noise at CES 2020
I think these are best apps out there for hearing impaired. Thanks, Abram for this beautiful post.
Sounds really interesting. On a sour note, I’m reminded of SnowOwl, who happily took my money and never even looked like delivering anything. That was a similar concept using dedicated hardware. It would have included equalisation to your audiogram and highly directional microphones as well.
This might come into its own in a year or two when we (hopefully) have a new Bluetooth audio streaming protocol with low latency. Right now, you’d really have to go wired, which would probably be a bit obvious.
Could that new streaming protocol sway the hearing aids with Bluetooth?
Well I think so, but I have no inside information. I posted a few links about it here: New Bluetooth audio standard on the way?. I’ve seen it mentioned in passing elsewhere but the Bluetooth Special Interest Group is not a leaking ship.
It seems they are developing a new audio streaming standard to replace a2dp. This protocol will support Bluetooth low energy audio streaming (immediate application for hearing aids right there), low latency (ditto), a single high quality codec, broadcast mode (so you could just choose to connect yourself to the sound source at your local cinema or lecture theatre), and true wireless stereo (so each ear bud or hearing aid receives its stream independently of the other- no need to communicate with each other through the wearer’s head).
That’s all I can glean and none of it is from ‘official’ announcements. Release was supposed to be about now, so who knows.
Edit: And just to state the obvious, even once they ratify and publish the standard it may take a long time before we see the product.