Re: about reading numbers in nvda 2017.2 and before: nvdadevelopers and experts please help me.
toggle quoted messageShow quoted text
It is very important to distinguish that NVDA is not, and I repeat not, responsible for the processing of numbers. All this functionality is controlled by the synthesizer of choice. In certain locales and regions, the period (or full stop) symbolizes the decimal point. After it is encountered, numbers are read by single digits by most synthesizers. Eloquence, with the American English setting, will pronounce the 0 as an ‘o’ as that is common in North America. I haven’t seen eSpeak do this any at all.
Sent from Mail for Windows 10
From: Chris Mullins
Sent: Thursday, August 3, 2017 4:28 PM
Subject: Re: [nvda] about reading numbers in nvda 2017.2 and before: nvdadevelopers and experts please help me.
Further to the topic of number processing, below are typical number sequences I come across on a daily basis and some observations regarding their processing by NVDA:
Decimal Number using thousands separator = 123,056,789.00
Speaks as one hundred twenty thre comma zero five six comma seven hundred eighty nine point o o
Decimal number without thousands separator = 123056789.00
Speaks as one hundred twenty three million fifty six thousand seven hundred eighty nine point o o
IP address = 192.168.20.1
Speaks as one hundred ninety two point one six eight point 2 o point one
Phone number 01246 567890
Speaks as zero one two four six five hundred sixty seven thousand eight hundred ninety
24 hour based time = 19:28:09
Speaks as nineteen colon twenty eight colon zero nine
* Digits prior to a thousands separator, space or line terminator are voiced as individual whole numbers except when the segment of the number begins with a zero in which case digits for that segment are spoken individually.
* Digits prior to the first period in a number sequence are spoken as a whole number, digits thereafter are spoken individually.
* After a period has been encountered in the number sequence, the zero digit is spoken is the letter o.
I don't know if the rendering of the letter o instead of zero is by accident or design when it occurs after a period in a number sequence but I consider it to be a bug. I have tried it with Eloquence and ESpeak and both work the same.
My keyboard and language settings are UK English for these tests. I don't know how Locale aware NVDA is as for example Germany use comma as the decimal separator and period as the thousands separator but it would be interesting to know if similar results to mine occur there.
From: firstname.lastname@example.org [mailto:email@example.com] On Behalf Of Quentin Christensen
You could get NVDA to read numbers how you want using a dictionary rule which uses regex. I think someone did it on this list previously. You would need to do it for each type of number, eg one digit followed by a full stop, followed by two digits, a full stop then more digits etc.
On Thu, Aug 3, 2017 at 5:08 PM, Gene <gsasner@...> wrote:
I don't know what NVDA does with different numeric reading settings. I just tried having it read a number with a decimal point with punctuation off using a version of NVDA that doesn't have the read number as digits feature. In other words it reads numbers the way NVDA always has. It reads numbers as thousand, hundred, etc. It says the word point where the decimal is and reads the number as a number before the decimal point and digits after the decimal point. For example, it would say 4 hundred forty three point 236. I'm not sure just what you want to hear.
As I recall what I saw on the list early in the discussion, the option to read numbers as digits is in a snapshot version. If it is in a snapshot branch where the program is expected to be stable, you might want to experiment. In the branch where the code is expected to be stable, snapshots usually work well. There may be a bug here and there but many people use snap shots from stable branches because they want newer features and stable operation. I've used a few snapshots from a stable branch and I've had no or almost no unexpected behavior. People who don't want to wait for production releases but who want a feature would do well to try a snapshot from a stable branch with the feature. There are three branches of snapshots. One or two or them is expected to have stable or quite stable code. I'll let those who follow snapshots and branches discuss the question further.
----- Original Message -----