Client Suggestion: Voice Recognition
I was driving from Las Vegas, NV to Fullerton, CA this past weekend. During the trip, I had a few hours to think of what client upgrades could be implemented into WAZE. Specifically, as the title of this thread suggests, I thought about Voice Recognition. This came about as I sent a report, but was unable to type a comment, thanks to texting and driving dangers.
Here is what I was thinking:
Voice Recognition would always be active while WAZE is active. Instead of pushing a button to activate Waze, what if a keyword was programmed to start the functionality. In this Example, the keyword is “WAZE.” Whenever the driver says this keyword, WAZE listens for the next word that is associated with the program. In the situation below, this is what it would be like to report a Police Trap.
In the simplest form:
“WAZE, Report, Police, Send.”
“WAZE, Report, Police, Comment, Hiding behind the tree, Send.”
Of course, after Report, this could be done with any of the report descriptions in the list. Consider also "Drive To" could be implemented into the system.
Currently available in English and in U.S and Canada.
Will be available to other countries as well soon.
elena chiriac commented
elena chiriac commented
Martin Baťo Bryndza commented
Why are people from other countries use voice commands for reporting of incidents even if they want to do so in English?
I have my phone in English, Waze is in english and voice navigation is in English. But still the voice commands option is hidden from me in the settings menu.
i think it should be like ok google, but instead ok waze
No touching the phone, just say
Ok waze , report car broken down on road ...
Jonathan Clutter commented
I really need this function to work better. I speak in a very clear, articulate voice with little to no discernible accent. And it often does not understand what I'm saying.
Though this could be caused by too much ambient noise or not having the phone close to my mouth, I have had trouble in a quiet car with the phone at my cheek. I suspect this is caused by a communication issue to the servers. In which case, I'd rather it respond that it couldn't communicate with the voice recognition servers so I know what the issue is, and not throwing my phone out the window in frustration.
Additionally, if Open Mic+ can do key word prompting, I'd like to have it as a switchable option. That way, people with battery life issues can turn it off. I, on the other hand, could be hands free, and plug my phone in to a charger to mitigate the additional drain.
I'm glad to see this feature will be available soon in other countries, I've just had the idea for it today. Glad to see others have thought it could be useful.
I'm not sure exactly how this option has been implemented, as it is not yet available where I live, but I think I know how I would like it to work. Having Waze listen for keywords is maybe not the best idea, because it drains precious battery (even a small amount), and it might not work very well if you have the car radio on (and perhaps loud).
Instead, similar to the previous suggestion of RhodesaSsassinScholar, shaking the iPhone — or even better — performing a screen gesture (tap twice with two fingers, or some sort of directional slide) to let Waze know that you want to report something, and then having Waze listen for key-words (like "accident" or "road hazard") would be way better, in my opinion.
Anyway, looking forward to whatever specific model has been implemented.
I work in asr and would be willing to help out with this.
This OUGHT NOT rely on a google service. I use my android without gmail and the phone performs swimmingly and longer. No google cloud / push for me.
Hotmail, whose owners license ActiveSync to google, powers calendar/contacts flawlessly
Voice recog ought be performed in non-google cloud. Waze seems to have preference for Amazon services, but for best results such technology ought be cloud-agnostic. There are dozens of tech competing to be open standards for cross cloud seamless functionality
I would prefer an option to have hand wave initiation instead of continous audio polling for trigger words.
Or perhaps a bit of shaking to initiate as some people might use arm rest or center console cup holder as makeshift "mount" for phone
Forks of CyanogenMod allow 90/180/270/360 auto rotation of screen content. In one vehicle I store the phone upside down in a cup holder to reduce hassle of power connector. It would be easy to shake the phone without taking my eyes off the road conditions.
Retasking a less used hard button as germaine suggests is also desirable -- I have no physical keyboard on my *phone*.
It's inane to pay additional fees to communicate is text. Especially paying to receive unwanted messages. I use XMPP. It furrher provides conversation encryption with OTR. I have to pay for data. A mature internet communication protocol is the way to go. No, gtalk isn't really xmpp: it's a failed gay attempt. Interweb search free public XMPP server to signup for a Jabber ID the cancel money burning sms. Your "free unlimited" texting with sms is a third or more of your bill. Soapboxing triggered by keyboard topic.
@op it is easier to train the user than to train voice recognition. While desirable voice recognition feature is masochism waiting to happen.
Instead tie voice monitoring to the camera button as not everyone travels in a silent car.
Continual monitoring would be a HUGE resource sink.
Vlingo has something similar - you say "Hey Vlingo" and it starts taking voice commands. I don't believe this particular "Magic word" aspect requires server offload.
In my country (Australia), touching a phone without having the handbrake on/being in "park" is illegal.
Most people don't follow this because it's a bit over the top (of course), but I'm not foolhardy enough to start navigating through menus at 60 km/h ;)
I suggest not to build speech recognition, and a kind of voice tags, which each in his own voice recorded for certain actions. In my old NOKIA was such a feature, I chose the name in the phone book and dictated his voice a voice tag, which is then compared with what I said into the microphone with voice dialing.
In the settings for the item VASES accident record voice tags "accident" and then when I need to mention the accident - an accident and I'm talking VASES comparing spoken sentences with a recorded voice tag that marks on the map crash. Then stopped, for example at traffic lights, I add - the direction of motion, etc.
I apologize for the Google translation
Robert, I was thinking exactly at the same improvement. And bpitas is on the right as well. Continuous speech recognition is a pain, while a single button to switch on the recognition would solve the problem in a satisfactory way. Hoping in this enhancement by the developers...
I work for a speech rec company. What you are asking for is called "Magic word" or "selective barge-in". The problem is, the good speech rec functionality, where it can transcribe what you say into a text field, is actually server based. The audio packets are captured on the handset and then sent to the server for recognition. So there is no good way to do "magic word" on a handset. Adding a button that does it is definitely doable though - plenty of our mobile clients do that.
It would be nice if avaliable in symbian phones too
Andrew Conkling commented
I think this would be the single biggest feature improvement waze could offer.
This is something really important and useful. This will improve, first of all, SAFETY. Also very important, it will improve the quality of the UPDATE REQUESTS as 90% of the requests lack detail, making them useless. I know this since I am Area Manager. Users and I waste lots of time with these useless requests.
I would like to be able to report things such as accidents to other Waze users without getting in to an accident while trying to do it. Please keep us posted on this one.
Martin Tlustos commented
Or you could put customizable buttons on the screen so you don't have to press four times but only once...
This already "just works" in Android.