
Hi. This is probably mostly for the webmaster, but I would like to get input from others on this list. Also, if you think this is off topic, I can move this discussion to the gutvol-p list instead. Has any progress been made with regards to making PG books available in Braille formatted files? There is at least one free Braille translator for *nix that I know of. I haven't seen this on the recode or bibrec pages. I would be willing to test the Braille output. I know of others who would probably also be interested. Probably a link on the recode page would be best, but it would be nice to see a link as part of the standard download links on the bibrec pages. That link could probably be a cgi program which would call the translator, NFBTrans, which would convert the file. I believe that it can be set to write to stdout, but it might be more convenient to have it create a separate file with the .brf extension for downloading. Here are a couple of issues to keep in mind. First, Braille only works with 7-bit, upper case files. NFBTrans will convert from mixed case just fine, but it is important that they are not 8-bit files. Braille does not support international accented characters. There are different Braille codes for other languages such as Spanish or German, but I don't know how well NFBTrans supports them. Besides it would be difficult to have an automated way of passing this information on the command line at the time of translation. Second, it will work only on plain text. It does not convert from html at all. Third, it might be best to have two links, one for a single file and one for multiple volumes. The reason why is that one printed page is about 3-6 Braille pages, depending on the size of the print and any illustrations. Many blind people now use special PDAs which can handle big files, but large files are bad for embossers. If someone downloads and tries to emboss a 300 page print book, it could easily be 900-1,000 pages in Braille. So, it is best to also make smaller files available. This is known as volumes. that same book could be downloaded in three Braille Volumes of 100-250 Braille pages each. That is much better for binding etc. I think the best way to do this is with the standard split utility. You can set it for X lines and that would be perfect. It would not indicate at the top of each new file that it is a subsequent volume in a set, but that can be added by the person doing the embossing if desired. I don't know exactly what the correct number of lines per file is, but I can check into that. Finally, after the translation to Braille options are set up and working, it would be nice if PG could have an official press release announcing this service. I know of several groups for the blind and many blind individuals who would be very interested. This is not as good of a solution as making books available in the new DAISY format, but I am not aware of a free html to DAISY converter, and I don't think that DAISY works on plain text at all. Once a master xml format is in place, conversion to DAISY would be even simpler, but I don't think there are any free tools. Really all that needs to be done is to integrate a script into the PG site that calls nfbtrans and generates the .brf file. The .brf can be sent to the browser for downloading. Windows users would probably have to right-click on the Braille links and save the files manually. It would not be much harder for the multiple volumes, just use a split utility on the .brf file and create a temporary page with each piece of the file linked so they can also be downloaded and saved. Does anyone have any thoughts on this? I don't mind testing or finding others to test, but I admit that my knowledge of cgi scripting is almost zero. I can compile nfbtrans easily enough, but I don't know how to integrate it into the recode facility or the PG site in general.

Tony Baechler wrote:
Really all that needs to be done is to integrate a script into the PG site that calls nfbtrans and generates the .brf file. The .brf can be sent to the browser for downloading. Windows users would probably have to right-click on the Braille links and save the files manually. It would not be much harder for the multiple volumes, just use a split utility on the .brf file and create a temporary page with each piece of the file linked so they can also be downloaded and saved.
Piping the files thru nfbtrans should pose no problem at all. Question: won't every blind computer user have this program on his PC already? Won't she be better able to tailor the output to her needs if the program is run locally? -- Marcello Perathoner webmaster@gutenberg.org

On Sat, Apr 02, 2005 at 10:32:36PM +0200, Marcello Perathoner wrote:
Tony Baechler wrote:
Really all that needs to be done is to integrate a script into the PG site that calls nfbtrans and generates the .brf file. The .brf can be sent to the browser for downloading. Windows users would probably have to right-click on the Braille links and save the files manually. It would not be much harder for the multiple volumes, just use a split utility on the .brf file and create a temporary page with each piece of the file linked so they can also be downloaded and saved.
Piping the files thru nfbtrans should pose no problem at all.
Question: won't every blind computer user have this program on his PC already? Won't she be better able to tailor the output to her needs if the program is run locally?
I think I corresponded with Tony about this a couple of years ago. I had no less than three programmers working on "conversion on the fly" which would generate formats including Braille, MP3, and others from .txt, .htm or .xml source files. Unfortunately, none of these ever became complete enough to offer on the Project Gutenberg download site. I am still very much able to provide a programming platform (a Linux server, with plenty of space and a copy of the PG collection) to people who might want to develop a CGI, PHP, Web services, or other platform for this type of functionality. Most of the tools already exist (i.e., for .txt to HTML, or Braille), but it's still a complex problem due to the complex nature of our collection. (That is, lots of different files & formats to choose from.) While Marcello is correct that many blind or visually impaired computer users already have nfbtrans (or something similar), I still think a general purpose conversion on-the-fly between formats is useful. And, if we offer this functionality, then an option to convert to Braille via nfbtrans is a very easy addition. There are just a few options for the output....like all of the other transformation programs... Long-timers on this list are getting tired of me talking about conversion on the fly, I know. Plus, this inevitably leads to a discussion of eBooks being "born as" XML, which I have also tried to facilitate. People who are newer (or newly-energized!) are welcome to look at viable methods for delivering some of this functionality. Beware, though: it's a larger and more complex problem than you might guess at first. Some folks will even remember that I offered a "bounty" reward for completely functional applications. This offer still stands. -- Greg

Greg Newby wrote:
I think I corresponded with Tony about this a couple of years ago. I had no less than three programmers working on "conversion on the fly" which would generate formats including Braille, MP3, and others from .txt, .htm or .xml source files.
Unfortunately, none of these ever became complete enough to offer on the Project Gutenberg download site.
This is a matter of a few hours ... its almost the same as the file recode service. I just need somebody to work out the nfbtrans command line options that work best for our etexts. Anybody got a braille embosser to test? -- Marcello Perathoner webmaster@gutenberg.org

Hi. I can help you with this. Please give me a detailed list of what you want/need tested and I will ask people to look into it. I know of at least two mailing lists that would have blind subscribers interested in looking at this. Remember that we need to keep in mind that many people will use PDAs designed for the blind. There is no real difference except that they don't need formfeeds while embossers do and generally one big file works better than multiple volumes. For embossers, formfeeds are a help, but nfbtrans puts that in automatically once you set the page length. You generally want the page length at 25 lines and 40 columns. If you can put files somewhere for people to download and test or somehow find a way for people to experiment with various kinds of output, that would be best. I think there might be someone on one of the lists that uses nfbtrans and could help more with the switches. Unfortunately, even among the blind it's mostly a Windows world so few people know how to use the command line. At 11:46 PM 4/3/2005 +0200, you wrote: This is a matter of a few hours ... its almost the same as the file recode service.
I just need somebody to work out the nfbtrans command line options that work best for our etexts. Anybody got a braille embosser to test?

Tony Baechler wrote:
I think there might be someone on one of the lists that uses nfbtrans and could help more with the switches.
Just ask some people to download a selection of PG files, run them thru nfbtrans and print a few test pages on the embosser or read them on the PDA. You should get at least a dozen or so different files from different producers done in different years. Formatting of PG texts has changed over the years. What we want is the set of commandline options that give the best results for most of the blind people out there. -- Marcello Perathoner webmaster@gutenberg.org

Hi Greg. If you corresponded with me, I have no recollection of it. I would like to see this happen, but I don't remember any discussion with you. I am the first to admit that I am not a programmer. I appreciate your offer, and it would be useful for compiling and testing nfbtrans, but I have no idea how to set up a script that would produce Braille files on the fly. Also, you said that it would be very hard to create Braille output because of the many formats involved. I couldn't disagree with you more on this. All Braille needs is plain text. Currently, nfbtrans won't work with other formats except text and a few language source code files. Simply pipe the 7-bit plain text through nfbtrans and don't worry about the format. My understanding is that even a master xml format will still have a 7-bit equivalent. For non-text, just don't offer a Braille option. There is probably a way to translate mathematics, but I am not aware of it with nfbtrans and it would add complexity because the .pdf or .tex would first need to be converted to 7-bit ASCII. I think that if nothing else, PG should offer Braille output just because the motto is that the files should be able to be viewed by anyone with any equipment. In many ways, literary Braille is similar to very old formats in that it is only plain text and is all upper case. Yet that is still what many blind people use, including the US Library of Congress. I have a question. How is it that you come to the conclusion that most blind people already have Braille translation software? I have read stats that no more than 12% of the blind can read at all. Of those, I would guess that not many have the computer knowledge to use a command line program such as nfbtrans. It is still a DOS-based program unless you want to compile the sources under Linux. I am not sure if development is still being done. I am not currently aware of an official download site but I can check into this if this is something that we're willing to move ahead on. I would be happy to look at the Linux server Greg is offering, but again I am no programmer so unless it can be done in a one or two line script, I'm rather helpless. Also, nfbtrans has many, many command line switches. It is designed to format Braille books, so offers facilities for running headers, table of contents, etc.

I can offer little commentary about Braille, but as far as audio formats are concerned, I can say that the visually impaired persons I have known who listen to substantial amounts of text often listen to it at substantially faster-than-intended speeds, and some do some other transformations on the audio to help them better catch meaning at these high speeds. Such things seem to be idiosyncratic to the individual listener. A fairly 'raw' format, such as .wav, would I think be useful to such users, but lossy compression formats have been engineered around assumptions about listening conditions that simply aren't true for the practices of those who listen to recorded speech as a way of life, and may be of little practical value. -- RS

Hi Robert. I agree with you and I am blind, so I'm glad you finally said it. I listen to text to speech at about 430 words per minute. That is slower than some people. I think it is great that PG has made some audio books available. However, to be blunt about it, I find them awful to listen to. They have low volume so are hard to hear and are very, very slow. I can't stand speech that slow! I know that the sighted public don't do well with computer speech, so I suppose it's necessary for them, but I would much rather have the speed set to at least 300 words per minute at the slowest. I wouldn't necessarily agree with you about raw wave files. I think mp3 is fine. One very big Internet radio station for the blind is ACB Radio. They are online at: http://www.acbradio.org/ They use mp3 exclusively, and at a rather low bitrate. I have no problem listening to it. In general, I don't like mp3. I collect old time radio and refuse to accept anything but raw wave files, but that is because it is of historical value and should be saved in the best audio condition possible. At 08:40 AM 4/4/2005 -0500, you wrote:
I can offer little commentary about Braille, but as far as audio formats are concerned, I can say that the visually impaired persons I have known who listen to substantial amounts of text often listen to it at substantially faster-than-intended speeds, and some do some other transformations on the audio to help them better catch meaning at these high speeds. Such things seem to be idiosyncratic to the individual listener.
A fairly 'raw' format, such as .wav, would I think be useful to such users, but lossy compression formats have been engineered around assumptions about listening conditions that simply aren't true for the practices of those who listen to recorded speech as a way of life, and may be of little practical value.

Hello. Surprisingly, many people do not have Braille translators. NFBTrans is the only free, open source translator I know of. Most cost many hundreds or thousands of dollars. In the case of PDAs, they can do translation but not reliably. There are many different types of Braille codes. There is Braille music, computer Braille, literary Braille and a special mathematics code. What we want is literary Braille since we are dealing with books. However, most PDAs and embossers would only output computer Braille which is harder to read and takes up more pages. If you look at a Braille embosser formatted file and compare it to the plain text, you will see that the Braille file is usually slightly smaller. That is because of contractions and other things to shorten words. For example, "sh" by itself in Braille is short for shall. "W" by itself is will. Unless such blind people are familiar with a command line or run Linux, most likely they wouldn't have access to a Braille translator and would probably appreciate Braille files being available directly from the PG download pages. I can query a few blind-specific mailing lists if you need more exact stats on how many people would be interested. It would certainly raise the reputation of PG if a press release was issued and the capability was implemented. At 10:32 PM 4/2/2005 +0200, you wrote:
Piping the files thru nfbtrans should pose no problem at all.
Question: won't every blind computer user have this program on his PC already? Won't she be better able to tailor the output to her needs if the program is run locally?
-- Marcello Perathoner webmaster@gutenberg.org
_______________________________________________ gutvol-d mailing list gutvol-d@lists.pglaf.org http://lists.pglaf.org/listinfo.cgi/gutvol-d
participants (4)
-
Greg Newby
-
Marcello Perathoner
-
Robert Shimmin
-
Tony Baechler