Test of Text-to-Speech (TTS) Abbreviations in Waze Clients

[sketch]

Agreed. The nuances of map editing are confusing enough already. We don't need more of them, especially just for pronunciation.

Every GPS system on the market pronounces street names wrong all the time. Waze does some better, some worse than others. Yes, we like to be better than everyone, but the benefit we'd gain from this is small compared to the pain of editing this into every segment of the map Waze might get wrong.

People can figure out that "Arnoo" means "Arnoult" and not "Severn" or "Hessmer" or any of the streets nearby. It isn't crucial that Waze says it with the T like everybody in town. It's good enough not to have to worry about it THIS much.


Sent from my iPhone using Tapatalk 2

[PhantomSoul]

I guess so, but I still think turning what is now a nice guided form, that I'm sure a lot of effort was put into making it that way, into a general command console is a bad idea. Besides this being a giant step backwards for where the WME system appears to be headed, I don't care what cone level/rank an editor is; he/she should not have to learn an entire set of cryptic (by cryptic, I mean anything not written in a natural human language - with spacing, punctuation, and capitalization that you would expect in normal narrative text of a human language) to control their input. At the end of the day, we're map editing volunteers - that have real professions outside of Waze - not some sort of system administrators.

Look, this is just my $0.02 for what it's worth. From my experience as a UI developer, there will always be people who want to bypass the UI development process with hacks like this to get quicker immediate returns. But in the long run, unless Waze creates some kind of administrator certification program to teach people these backdoor hacks - and limits such editing to only those people, it will create confusion, which will compromise the accuracy, reliability, and ultimately the overall usefulness of the data, reducing it to just meaningless clutter in the database, something that many other conversation threads, as well as established editing policies clearly suggest is a major concern.

[berestovskyy]

Frankly, introducing a hash tag syntax to names just further complicates a UI that already appears overwhelming for many new editors.

So we have suggested it works/appears for 3 or even 5 cones editors only. Also we can use Key-Value pairs "just like Alt names", not abusing alt names.

A bigger caveat to this, however, might be how many editors are familiar with universal phonetic spelling, and how many of those are familiar with how to produce many of its whacky characters on a keyboard?

In fact, in Nuance it looks like this:

b a k w *r d z -> backwards

Here is the Nuance Grammar Developer's Guide (PDF), page 85.

A separate TTS field would encourage people to put there something. On the other hand, a Key-Value pairs would allow to fix TTS/shields etc for experienced editors only. And we know, the majority of the segments do not need any TTS correction or shields at all.

[PhantomSoul]

Frankly, introducing a hash tag syntax to names just further complicates a UI that already appears overwhelming for many new editors. Further, from my understanding, Twitter-style hash tags are intended for indexing hot words and related topics, not for altering the way input is handled. The WME is a guided input form that instructs you to make specific inputs for specific purposes, not a general command prompt.

If phonetic names are that important to have different behavior from the alternate street names (which appear to be intended for a future address resolution system; no TTS or banner display involved) then we should get Waze to add a text control that actually prompts for phonetic input. A bigger caveat to this, however, might be how many editors are familiar with universal phonetic spelling, and how many of those are familiar with how to produce many of its whacky characters on a keyboard?

[vectorspace]

The spreadsheet lists TR-xxx as a proposed abbreviation for Township Road.

Thanks. I looked at the spreadsheet to check but missed that. If so, I'll seek an alternative abbreviation with BIA.

[mapcat]

(2) TR-xxx

Pronounced "Tribal Route" followed by the characters (up to six, but could just be one). This would be a generic tribal route designation for any tribal land that would accommodate it. This is not intended to be mandatory for use on all tribal lands, but rather a common form similar to "SR-xxx" for state routes.

The spreadsheet lists TR-xxx as a proposed abbreviation for Township Road.

[vectorspace]

From a number of visits with government agencies responsible for roads (see other threads) I have a few requests for addition to the TTS system.

(1) BIA-xxxxxx

Pronounced "B.I.A" then the following characters which will usually be a number, but could be a letter followed by a number in the future. This comes from a visit with Dept of Interior, Bureau of Indian Affairs, Supervisory Road Engineer.

(2) TR-xxx

Pronounced "Tribal Route" followed by the characters (up to six, but could just be one). This would be a generic tribal route designation for any tribal land that would accommodate it. This is not intended to be mandatory for use on all tribal lands, but rather a common form similar to "SR-xxx" for state routes.

(3) FSR-xxx

Pronounced "Forest Service Route" followed by a set of characters that would name the road, typically numbers and letters. This comes from talking to a US Forest Service GIS specialist that will be providing us forest road data, first for a few forests, and if that works we can have the forest data for the whole US.

[gerben]

Well, if they implement a filter for #tts, maybe it is better to implement a dedicated tts input field.

[berestovskyy]

Working with Waze to fix this is the way to go.

Sure! This is just a suggestion

This would be a very bad idea and would defeat the true purpose of Alt Name fields (which is for search results).

No, it's not. For instance:
Street: Northeast Kentucky Industrial Parkway <- this would be used by default (for search in this case)
Alt City: #short Alt Name: Northeast Pkwy <- this would be used to display and for TTS

Correction of pronunciation:
Street: John Paul IV Street <- this would be used by default (for search and to display)
Alt City: #tts Alt Name: John Paul 4th St <- this would be used for TTS

The #tts name could be a transcription (phonetic spelling) of the street name.

Having a prefix (#) it's easy to disting those properties and not use them for search if you like. But its not such a bad idea if one could find both '4th' and 'IV' IMO.

However, it wouldn't be a bad idea for a new dedicated field in which a phonetic spelling of names could be entered.

You see! They call it a brainstorming

[kentsmith9]

This conversation has been launched in two separate threads. This one and viewtopic.php?f=8&t=47255&start=10#p440213