Bluetooth and HFL
Bluetooth and HFL
The RL is my third Acura- and the third one with Bluetooth and HFL. Is anyone able to give a telephonic voice command and have the onboard computer pick the correct name every time. I must say that I have about 30 numbers loaded and some of the names are similar, but some aren't even close and HFL will select the wrong name at times. All 3 of my cars have had this problem although after 2-3 tries it always gets it right. Other times it works flawlessly. Screws up a demonstration to my buddies sometimes!
My wife thinks I don't always annunciate properly, I feel that the technology isn't where it needs to be to always work properly. What say you? <any O'Reilly fans here?>
My wife thinks I don't always annunciate properly, I feel that the technology isn't where it needs to be to always work properly. What say you? <any O'Reilly fans here?>
No major issues here. THe only time it get's confused is when I have two similar sounding names.
i.e. "Call Mom's mobile" vs. "Call Mark's mobile". I would say it get's my commands right about 95% of the time.
i.e. "Call Mom's mobile" vs. "Call Mark's mobile". I would say it get's my commands right about 95% of the time.
To be fair & balanced, Voice Recognition technology can never be fool proof.
First, it is looking for common denominators in speech, tone & inflection. Linguistics are tricky to dilineate. Each person has unique tone and inflection, and even the same person has variations. Few people actually focus on their personal speech patterns and project thier voice the same way when being 'recorded' as opposed to typical speech & conversation.
Next, the vehicle is not a soundproof environment and the microphone is not upon our person. Likely you recorded the HFL voice tags while the vehicle was idle (around 42 db). When driving you may be around 60 db, so already there is additional sound the HFL must process through. Add in road noise, tire thumps or any other ambient 'noise' and you are challenging the HFL Voice Recognition Technology.
I actually find the HFL in the RL much better than the TL. But I suspect that may be due to the RLs quieter cabin. I am in fact, very impressed what the VRT can really do. It is apparently better than I. I am often reminded by women that the car listens better than I do.
Anyone else care to opine?
First, it is looking for common denominators in speech, tone & inflection. Linguistics are tricky to dilineate. Each person has unique tone and inflection, and even the same person has variations. Few people actually focus on their personal speech patterns and project thier voice the same way when being 'recorded' as opposed to typical speech & conversation.
Next, the vehicle is not a soundproof environment and the microphone is not upon our person. Likely you recorded the HFL voice tags while the vehicle was idle (around 42 db). When driving you may be around 60 db, so already there is additional sound the HFL must process through. Add in road noise, tire thumps or any other ambient 'noise' and you are challenging the HFL Voice Recognition Technology.
I actually find the HFL in the RL much better than the TL. But I suspect that may be due to the RLs quieter cabin. I am in fact, very impressed what the VRT can really do. It is apparently better than I. I am often reminded by women that the car listens better than I do.

Anyone else care to opine?
I had a lot of trouble getting HFL & Bluetooth to understand me until I learned a trick from this forum. Don't raise your voice! When I stopped trying hard to enunciate and be loud, the system started to understand me. Completely counter-intuitive, but it worked for me. So just relax and see if it helps you.
Originally Posted by scottj
I had a lot of trouble getting HFL & Bluetooth to understand me until I learned a trick from this forum. Don't raise your voice! When I stopped trying hard to enunciate and be loud, the system started to understand me. Completely counter-intuitive, but it worked for me. So just relax and see if it helps you.
When you think about how the tech works, all it's doing is comparing bits. You record the name, and the system digitizes the recording. It samples it in a few places to come up with a "signature" print on the sample.
Then it listens for voice tags, and compares them with recorded samples.
The sample quality isn't great, i'm sure, because there's a big tradeoff between accuracy and speed... and further, things like vehicle noise, fan speed, external noise, etc. always figure into it, and nobody ever says anything exactly the same way twice.
So, it can't be perfect. But, you can make it work better by not recording similar names, and using more than one syllable.
I have a "Buck" and a "Seth" on my system. They always get confused. But "Joe So-And-So" always works.
The only time my system does not recognize the name is when my wife or other passengers talk (which is almost constant), then the VRT comes back with PARDON. I think it is otherwise remarkably accurate. Ah the joys of driving alone.
Thread
Thread Starter
Forum
Replies
Last Post
MetalGearTypeS
3G TL Audio, Bluetooth, Electronics & Navigation
6
Aug 29, 2016 08:28 PM
InFaMouSLink
Car Parts for Sale
3
Oct 30, 2015 09:43 AM
SUPRMN84
3G TL Audio, Bluetooth, Electronics & Navigation
5
Oct 7, 2015 09:46 PM
phillyguerrilla
3G TL Audio, Bluetooth, Electronics & Navigation
0
Sep 26, 2015 11:27 AM



