(old)Puppy Linux Discussion Forum

project homepage: http://espeak.sourceforge.net/


This pet package installs:

espeak
espeakedit
mbrola voice en1


At first I wanted to create a pet just for espeakedit, but I saw that previous versions of espeak didn't work with 1.42.04. This is the reason why I have added espeak too. If you are looking for espeak only click here

The aim of espeakedit is to modify the stress in words that may be spoken in the wrong way by espeak or create new words and phonetic rules. If you are game, you may even add a new language altogether. Feel free to share your improvements if you like.

I have put in the package the mbrola voice en1 to easily enjoy its very good quality. If you look for other mbrola voices launch espeakedit, select help, see instructions about mbrola.

Also, in the pet you can find a small script for idlers, which reads txt and html files with mbrola voices without typing long commands. However, remember only en1 is present after download. As a bonus the script may convert txt and html files into mp3. It uses lame, which should be installed in all pups equal and above 4.0 series. To launch type eread in a terminal.

At last a quick start guide is placed in /root/my-documents and an entry is added to the start menu.

tested in 4.2.0, 4.3.1, 5.0 (lupu) but it may work in other versions too.
To stay on the safe side if you have previous versions of espeak uninstall them before installing this package.


download espeakedit (21396145 bytes)


checksum: a6275ecbf2fb4ab14480d082c4b4aa98

Tried in 4.31


The ereader worked OK from the command line
but no sound from the editor

Anyone else have better luck?

I been working with TTS for about four years. I commute 75min one way to work and use TTS to create audio books. I think projects like this are great. I use Cepstal David for my TTS needs. $30. This is the only piece of software I have purchased for my computer in four years as well. I'd love to help your project and see TTS as a great resource that is very understood and appreciated. -Mark

hi lobster

Oops! I have the pc so crammed with so many libraries that I've forgot some. Thanks for your report All those who have already downloaded the pet should install the patch, anyway.

I'll upload the completely fixed pet as soon as possible.

I've testet the package with 4.3.1 and it's all right now.

@ mwestlund

hi mark, I must say that I am not in the espeak project, I wish I was. I'm only a supporter that would like to contribute to spread this great application as far as possible because I find it useful. And I think that if more and more people get involved then we will see espeak improve faster. At the same time puppy is as great a project and so you may have two instead of just one. not bad at all.

I'm glad that you're willing to contribute to espeak and I hope many other puppians will follow.

cheers

...in the meantime i've found out that the script eread had some problems too, even after the patch. now it's fixed.

later on i'll upload a single pet with all things running properly but for now everybody should download patch and ereadpatch separately.


ereadpatch

Hi there - many thanks

added patch
ereader will now read a text file
but still not speak what is in the top left speak box
What am I dong wrong?

unfortunately you are not doing anything wrong. the fact is that I left out some more libraries

this because I had some trailing libs that made all work for me. now I have tested espeakedit with new patch in a brand new fs and it should be fine with new patch. in case it's not let me know...

anyway you will see that `speak` should be used after `translate`

Everybody is encouraged to download the new patch which should be added to the old one. here you can also find an updated eread script that will place all the converted mp3 files in the current directory instead of leaving them scattered in the subdirectories. you can unistall the previuos ereadpatch, if you have installed it before.

if no other bug will be reported I will upload a single new pet soon

new pet uploaded in the first post. it merges all previous patches. in addition the script eread now reads and converts html other than txt files.

here attached the new script for those who don't want to download all the package again.

I've uploaded an enhanced version of eread. I have added several functions. Well, what we've got here is something you'll love it or hate it. I've started to build a tiny CLI application that could simplify the use of mbrola voices and ended up creating a small file browser. Many would say that there was no need of another CLI browser and nobody would use it. But what I actually had in mind was a tool for reading and converting txt files. I just wanted a fast CLI way to move through directories with easy access to different voices/languages.

The result is this version of eread. Not only can it read and convert text files easily into mp3, but it can also open a terminal in remote paths that you have even forgot that are there. It can also:

-open images
-'read' documents. That is to say, open docs with the suitable application (Open Office)
-'read' music (play files with gxine)

all this without leaving the keyboard for the mouse. This is the reason why I find it so fast.

And you may add whatever extension or application to open if you modify the script. It's not a final version. Any comments are welcome. If anybody is interested I'll write some documentations. For now you can find a help menu in the navigation menu (the one with the green-colored path) typing H (uppercase) and enter.


Last but not least. At some point I realized that I had in front of me a tiny but usable CLI-file browser. And what if instead of printing the output on the screen we redirect it to espeak? I thought that the blind can navigate the content of their directories with a simple CLI browser of few kilobytes and launch music files, read ebooks or open whatever application is in their computer. In fact, every file that should be printed to the screen is associated to a number that is read by eread (see screenshot). When you have listened to the number associated to the file you are interested in, you just press CTRL + C to stop espeak from reading the list and then type that number. It should be fast with the numeric pad of the keybord. Furthermore, you can search in directories jammed with files with a search function that allows to read every file that matches the string you have chosen to search. Three characters most of the time are enough. And to browse faster you can store to a file the current directory to retreive it later with ease.

Only one thing. To have a positive experience from the reading of the list of files, with espeakedit you should change something in the en_list file (or your language_list file for any other language):

# haS $max3

should be turned into

# ko@-*@sp'Ond,tu: $pause

the phonetic translation of 'correspond to'. It have to be associated to the graphic character '#'. For other languages the translation is obviously different.

And to each number you should add $pause

And then compile the changes.


I don't know if this could be really useful. I've never used applications like festival and I can't make a comparison. I need the opinion of some testers on this. If it is worth for anybody I'll give it a further go to improve it.

Notes for users.
-eread creates a directory named eread-converted in /root/my-documents where all the converted files will be put after conversion.

-a hidden directory named .eread is installed in /root where some files needed to operate are being stored.

-as I am still tinkering with the script, to open a non-txt file you should 'read' it with read option r. If it doesn't work choose the file from the 'lower directory' option L (lowercase). If it still doesn't open then the extension is not inserted. You can do it manually adding it in the script.

new build.

eread 0.7.1 for the blind. less chatty and then faster.

-interface speaks with mbrola now
-more convenient navigation
-you can read a file from the l option with current language
-some bugs are fixed

Nice work -but are all those TABS in the script necessary -I mean do the affect the output in some way?
Does this require any changes to the espeak sources?

Nice work -but are all those TABS in the script necessary -I mean do the affect the output in some way?

hi there, thanks. I hope it could be useful to somebody, anyway.

in my personal view tabs help give a better understanding of the script but don't affect it's execution in any manner.

Does this require any changes to the espeak sources?

no. espeak remains untouched. I have only suggested in one of my posts to modify the rules of the pronunciation of the hush symbol # with espeakedit. but to do this you just need to click on a button of the gui of espeakedit named compile. that's all. it takes less than one second. but this is not even necessary. it should only result in a less mechanical voice when reading long lists of files. you may even change # in a pause if you want, or leave it unchanged.

the output of the version for the blind, at the moment is messy, but only because the script is a work in progress and I want to see what's going on. of course it could be cleansed.

by the way, I forgot to say that in the head of the script you can easily modify the speed of the reading of items.

Oh, I agree that indenting is essential to readability in good coding, but here the above comes out with *twenty* TABS of spacing which means I must scroll horizontally to read anything that follows. What crazy editor are you using?

this crazy editor is geany with automatic indentation. like I said, I used tabs because in *my view* the script is more readable

2017, that works well ! i will improve my english..
article en français

çà fonctionne. The speaker speaks French but is born in Japan, no doubt about that. French can understand
you can find better, but you can find worse. Can help to learn English, really. Pronunciation will always be better than ours (french)

but my colleagues have really no interest in improving English