Copyright © 2003, 2005, 2006 Robert J. Chassell
In this essay, I suggest how to introduce Emacspeak to someone new, and I discuss some of the notions behind GNU Emacs, which underlies Emacspeak.
In addition, the essay contains some comments on two software text-to-speech synthesizers and one method of installation.
Last updated on: 2006 Mar 18
This essay uses a verbatim license, since it is a statement of my recommendations.
Permission is granted to make and distribute verbatim copies of this entire document without royalty provided the copyright notice and this permission notice are preserved on all copies.
This essay is about Emacspeak and how to introduce it to a friend. It is not a substitute for the Emacspeak manual, but an addendum.
In this essay, I will recommend that you and your friend read parts of the Emacspeak manual together, practice commands, and practice using the online help. In addition, I shall talk about the notions behind GNU Emacs, which underlies Emacspeak.
Your friend may be permanently blind, partially blind, or like a car driver, situationally blind. Regardless of circumstance, your friend will find that Emacspeak is different from other user interfaces.
You will notice that I keep mentioning `your friend'; of course, you may be introducing yourself to Emacspeak. This is not impossible. It is just harder. This is how I learned Emacspeak.
The essay has six parts. The first four parts are the core. The fifth and sixth parts are about particular software text-to-speech synthesizers and about one method of installation.
A short introductory segment, for orienting yourself. I ask two rhetorical questions.
This is the longest part.
How to download texts, such as Alice in Wonderland, from the Gutenberg Project.
I cannot imagine running Emacspeak without customizing it.
A discussion of eflite and espeakf. When I first wrote
this in April 2003, I had problems with them. Now, in
March 2006, they are good enough.
This is a section on installing Emacspeak in Debian with both
eflite and espeakf. Even though I am now running two
instances of Emacspeak, one using eflite and the other using
espeakf, I am least certain about this part. This is because
so much of my Emacs and other installation is from sources rather than
from standard, precompiled binaries. I don't know what a standard
Debian or standard Emacspeak installation involves. I think what I
describe is correct; but am not sure.
My first question is whether you already know and use Emacspeak and are introducing someone else to an already running Emacspeak with which you have experience, or whether you are the new user and must not only learn to use Emacspeak yourself, but must also install it? (I know there are other alternatives, but these are critical.)
My hope is that you are the first: that you, the reader, are someone who already knows and uses Emacspeak and are introducing it to someone else; but my fear is that you are the second. The second path is harder.
I will mention installation again after introducing Emacspeak, talking about some of the underlying notions behind GNU Emacs, and several other matters. But first, I would like to ask you what kind of text-to-speech facility you have?
Are you using an external synthesizer, an internal hardware card for
text-to-speech generation or do you expect to use a software
text-to-speech package such as eflite or espeakf? I do
not have a hard synthesizer, but run both those two software packages.
In the installation section, I will talk about them.
Emacspeak was originally written for an external hardware DecTalk synthesizer attached to a serial port. I am told that a hardware synthesizer is better, whether internal or external, and that if you can, you should use one. However, you can also use a software text-to-speech program.
This means that you can run Emacspeak on any computer that has sound,
as do most contemporary personal computers. You can try out Emacspeak
without first investing in special hardware. However, it is also
worth bearing in mind that at the time I am writing this, in May 2005,
eflite and espeakf have different virtues. Neither is
as good as it might be. For example, my installation of
espeakf sometimes restarts its speaking of punctuation after I
shift back to a buffer, but not always. On the other hand
espeakf has voices, which I really like. All in all, the two
software synthesizers are good enough.
Let me hope that you, the reader, already know and use Emacspeak and are introducing it to someone else. You may not be that person, but I will write this section as if you are. You are introducing a friend to Emacspeak.
First of all, in Info, please listen together with your friend to the beginnings of the Emacspeak manual. You might want to start with the Introduction node. It is well written. See Introduction. You can use the opportunity to introduce various Emacspeak commands and to discuss the philosophy or world-view behind Emacspeak.
You can also explain the meaning of some of the jargon, such as `point' to mean the place in the text where commands operate, `visit' to mean opening a file in a buffer, and `string' to mean a `string of characters'.
Sighted people tend to work with the three contemporary user interfaces: command line, graphical, and Emacs. Each is different.
A key to Emacspeak is to understand that it provides a fourth type of user interface.
Emacspeak is different from the shell command line interface that uses commands such as `ls'. It is different from a graphic user interface such as GNOME/sawfish that uses a package such as `Mozilla'. Moreover, although it is based on the virtual lisp machine interface that is GNU Emacs, it is different from it, too.
Put another way, Emacs and Emacspeak are both integrated user interfaces, like a shell, plus `vi', plus `ls', plus `gdb', plus `gcc', and so on. While related to each other, they are also different.
Emacspeak is a complete audio desktop; it is not a visual desktop. T. V. Raman describes it in his Emacspeak manual. See Emacspeak.
Your friend will want to learn a great many keystroke commands, but if he is like me, he will continually forget them, too.
So you need to teach him how to use Info, which is where he can listen to the Emacspeak manual. And you need to teach him how to use help. This way, he can always learn more, or relearn what he has forgot.
The first commands I learned were:
Control-h Control-eI use this command all the time, to remind me of other commands.
(In most Emacs documentation, this command, Control-h Control-e
is written as C-h C-e.)
Control-e left-braceWith a prefix arg, speak the rest of the current paragraph. With a
negative prefix arg, read from the start of the current paragraph to
the current position, that is, to point.
Control-uUse as a prefix to some interactive commands, such as Control-u Control-e left-brace, which speaks the rest of the current paragraph.
To read from the beginning of the current paragraph to point,
type Control-u hyphen one Control-e left-brace.
Control-e left-angle-bracketWith a prefix arg, speak the rest of the current page.
I often use the Control-u Control-e left-angle-bracket command.
Control-e sControl-h iIn Info, press h for an Info tutorial.
Control-h tYour friend will also need to learn the regular movement and search keys. The Emacs Tutorial is good for this. But first teach him how to listen to Info and to use the online help.
Note that the Emacs Tutorial contains left and right angle brackets in two places. To hear them in Emacspeak, you must set the punctuation mode to all, with the command:
Control-e d p all RET
Otherwise, the brackets will not be spoken. (The <RET> at the end of the command means to press the Return key after typing all.)
In most Emacs documentation, a command such as Control-e left-brace
command is written as C-e {. If you are listening in
Emacspeak, you have to set the punctuation mode to all;
otherwise, you will not hear the left-brace spoken.
As for movement: incremental search is the most important way of moving around documents. This command is bound to Control-s, which is usually written as C-s. Often people do not think of a search command as a movement command, but it is.
Moreover, Emacs incremental search provides the single best user interface for a search command that exists. Many other search commands require that you figure out your search string before you type, which means that the search fails when you type too little. Incremental search is nicer to use.
The character and word movement keys, and the character and word deletion keys, are reasonably easy for English speakers to learn, since they use English language mnemonics: Control-f to move forward a character, Meta-f to move forward a word. (Meta-f is usually written as M-f.) p is for previous, n is for next, b is for backwards, and d is for delete.
Control key commands are often (but not always) for character movement; Meta key commands are often (but not always) for word movement.
You will want to explain the use of Meta. You might tell your friend that on many modern keyboards, a Meta key is what is now, generally and erroneously, labeled Alt, spelled A L T, and usually written <ALT>. (Genuine Alt keys can and do exist; but the default keybinding for many machines automatically makes the key labeled Alt be a Meta key. Naturally, you should set up the machine so that a key labeled Alt takes on a Meta keybinding, for example with the install-keymap emacs2 command.)
Worse, in some cases, people press the Escape key to simulate a Meta keypress. The Escape key is usually written <ESC>. No one in their right mind uses an Escape key, unless forced by a really old keyboard.
While I am speaking of keybindings, and to be complete, please be sure the control key is the key to the left of the A key. This is something you should set up; you don't need to explain any of this.
You may have to change the keybinding. On some keyboards, the key to the left of the A key is labeled Caps_Lock. It is as if the keyboard manufacturers think that computers are typewriters from 1885. Moreover, they put the key labeled Control in an awkward spot, and then provide default keybindings that match the labels. You have to change the keybindings.
Generally, the proper keymap is in
/usr/share/keymaps/i386/qwerty/emacs2.kmap.gz
or
/usr/share/kbd/keymaps/i386/qwerty/emacs2.kmap.gz
You can install the keymap permanently on your system for a console with the shell command:
install-keymap emacs2
(This may be a Debian specific command; it copies the emacs2.kmap.gz file to the /etc/console/boottime.kmap.gz file, which is loaded into the kernel at boot-time. The older and widely used command is loadkeys; when you run it, you have to specify the full path to the file.
(For example, to install the proper keymap, you might type
loadkeys /usr/share/keymaps/i386/qwerty/emacs2.kmap.gz
(Note that since files in the /usr/ directory are sometimes not available early during the boot process, it is a good idea to copy the /usr/share/keymaps/i386/qwerty/emacs2.kmap.gz to the /etc/console/boottime.kmap.gz file. This is what the install-keymap command does. If you don't have an install-keymap command, you can copy the file yourself and put a loadkeys /etc/console/boottime.kmap.gz command into one of your boot scripts.)
If you are using an X Windows graphical user interface, you may need to specify the keymap that X uses in addition to specifying the keymap used by a console.
You can put the following commands in your ~/.xsession or
~/.xinitrc file:
xmodmap -e "clear Lock"
xmodmap -e "add Control = Caps_Lock"
Some systems reverse the meaning of the Help key, which is Control-h, and the Delete key. The node Keyboard Translations in the Emacs manual describes what to do.
You can go directly to that node in Info by evaluating the following expression, that is to say, by positioning point after the final parenthesis and typing Control-x Control-e, which is usually written as C-x C-e:
(info "(emacs)Keyboard Translations")
(Please remember to note my use of the jargon word `point'; it needs to be explained as the place in the text where editing commands operate.)
Returning to what you might talk about early on with your friend:
He may want to set the voice synthesizer speech rate. I use both the dtk-set-predefined-speech-rate command:
Control-e d 1
Control-e d 2
Control-e d 3
and the dtk-set-rate command, such as
Control-e d r 230 RET
Probably it is a good idea to pick one of these commands. Two sets of commands may overload his memory.
As he gains more experience, your friend will be able to understand text spoken faster and faster, and will want to increase its speed.
Note that in both set-rate commands, an interactive prefix means to set the speech rate globally. Otherwise, it is set only for the current buffer.
Your friend may want to set rates globally, so it is worth telling him about typing Control-u first.
For a description of the basic commands, see the node `Reading' in the Info file emacspeak.info
(info "(emacspeak)Reading")
(As I said earlier, you can go directly to that node by evaluating the Emacs Lisp expression: position point after the final parenthesis and type Control-x Control-e.)
Perhaps evaluation is too much too soon, since it means remembering a new command as well as figuring out where `point' is. On the other hand, I think it is worth illustrating the action, since it opens up future possibilities.
Incidentally, you can go directly to the Introduction node of the Emacspeak manual by evaluating:
(info "(emacspeak)Introduction")
My hope is that you will inspire your friend to experiment with many commands and learn a few of them; you will inspire him to learn how to use the help features, to listen to the Emacspeak Info manual, and then to listen to the Emacs manual.
After that, he can start using Emacs W3 or Emacs W3M mode to browse the Web, Emacs RMAIL mode to listen to Email, and Emacs mail mode to send it. You may need to make sure his machine is properly configured for all this.
While listening to the Emacspeak Info manual, it is worth explaining some of the problems people have with the Emacs manual, to which he should listen next, or at least listen to part of it. I do not recommend listening to it all at once.
The Emacs manual is tightly written. When I first read it, I found I sometimes had to spend several minutes puzzling out a concise meaning. The good news is that it is accurate and well written. And after you understand it, the sentences make perfect sense.
When I first started, I read the manual in segments over 2 years. (This was in the 1980s.) I did not understand all that I could do at first. Indeed, for the longest time, I did not switch from using `ls -al' in a command line to using Dired mode.
Some people say that the Emacs learning curve is steep.
In my experience, this is not true. It took me about five minutes to start to use Emacs productively. But I could not and did not do much at first, except write.
However, the learning curve will be steep if you try all at once to use many of the features that Emacs provides. I have described GNU Emacs as
... like having a dragon's cave of treasures.
You should not try to be too greedy and over-stuff yourself all one sitting.
(That quotation is from the Info node "On Reading this Text" in the Info manual that introduces programming in Emacs Lisp:
(info "(eintr)On Reading this Text")
I wrote that text "as an elementary introduction for people who are not programmers", but presupposed experience with Emacs. So I doubt you or your friend will want to look at it early on. But I hope that he will keep it in mind for the future.)
To listen to the Emacs manual in Emacspeak, you will need to set the punctuation mode to all with the command:
Control-e d p all RET
RMS, Richard M. Stallman, wrote the Emacs manual.
Unfortunately, RMS thinks of editing as having a very large meaning. He writes as if using ls and rm in a command line is the shell equivalent of editing in Emacspeak. People who use a shell do not think of this as editing.
However, if you think about it, you will discover that the Emacspeak actions in Dired mode are a kind of editing. Indeed, the name Dired is a contraction for the phrase `directory editor'.
But the kind of editing that goes on in Dired is not same kind of editing as people do in an editor like vi when they press the h, j, k, and l keys to move around. People are fooled by the language.
The words mode and library are not confusing, but may not be understood immediately. The two provide the features that programs have in other interfaces. Thus Dired mode provides the same features in Emacs that the combination of ls, mv, chown, chgrp, gzip, more, vi, and rm provide in a shell. A library provides a mode.
Because of the adoption of Xerox Parc style windowing systems over the past generation, many people are confused by Emacs' use of the words window and frame. Emacs provides for multiple windows within a frame, as it always has.
You will probably not use a graphic user interface at all; but enough people do that you will hear what they say.
Nowadays, sighted people often think of a window on a computer screen as being a contiguous, usually rectangular space, what in Emacs is called a frame. That is because Emacs was designed initially to fill a complete display as a tiling window manager. (A tiling window manager is one in which windows do not overlap, but are contiguous, like physical tiles.) Parts of the display were called windows because they enabled a sighted person to look at all or part of a buffer.
Companies like Apple and Sun, and the X Consortium, copied Emacs jargon for their own windows, to mean a part of a screen. (Or else the notion of a window was generic and commonplace.)
Thus, the term window started out and continues to mean a part of a display.
But when Apple, Sun, and the X Consortium, and their followers, adopted the term window to mean a part of a display, they lacked a term to handle a part of a window. (The word `pane' was suggested, but never popular, because it sounds similar to the word `pain', spelled p a i n, and because some people were accustomed to material windows that were not made up of multiple `panes of glass'.)
However, in those days, mostly the 1980s, windows seldom contained parts, other than a menu or tool bar or panel that applied to a whole window. Thus, one spreadsheet would appear in one window; one file would appear in another.
Indeed, many of the non-Emacs programs I use today in a graphic user interface still tend to put one set of contents, with its associated panels and tool bars, into one window: for example, XMMS, Mozilla, or gnome-apt.
With Emacs, on the other hand, you could always put a directory listing, two files, and an email message into four different parts of an Apple or X style window (although most people keep to one or two parts most of the time). These different parts had always been themselves called windows, and so they remained. Hence, the invention of the term frame to refer to a segment of a display as produced by an X or Sun user interface program.
Nowadays, you can start different frames in an instance of Emacspeak that is not running in a graphical user interface. One frame overlies another. The mode line will tell you which frame you are in. (Type Control-e m to listen to the mode line. That key chord calls the emacspeak-speak-mode-line command. The frame number is listed in the second and subsequent frames.)
Emacs jargon can be brutal. Some people are upset by the use of the word `kill' to mean `cut' as in `cut and paste'. (The word `cut' also upsets some people, but generally less than the word `kill'.) In Emacs, an entity `killed' can be resurrected, which I think of as a Christian form of meaning.
Some years ago, at the transition from Emacs version 18 to version 19,
I offered to convert every use in Emacs of the word kill to
clip. There were more than 400 such uses. You could clip a
segment of text, and if it were never put back, it would be deleted.
Alternatively, you could yank the segment from the clip-ring.
I picked clip as the replacement word since it has four letters
and fits the older formatting and language without requiring much
rewriting.
Moreover, had I changed the term to `cut', people would have expected
me to replace the word yank with `paste', to fit the phrase
`cut and paste'. But I do not like the word `paste'; I no longer
literally paste clips onto a sheet of paper as I did when I was young.
It makes more sense to me to yank a segment back from the
clip-ring.
However, RMS never wanted me to make the replacement. I think part of the reason is that at that time, RMS still enjoyed what I think of as a dead joke: the manual for version 18 Emacs said:
... you don't have to kill all the text in one command; you can keep killing line after line, or word after word, until you have killed it all, and you can still get it all back at once. (Thus we join television in leading people to kill thoughtlessly.)
(The parenthetical remark about television was removed sometime later and the other wording slightly improved.)
Perhaps more significantly, RMS asked many people whether they were
bothered by the jargon use of kill. They said `no'. It turned out
that RMS asked people he thought of as significant, namely people at
M.I.T. I think that was too narrow a group. But that is what
happened, and I still think it was a mistake.
Most important of all, please emphasize and explain that Emacspeak is safer than many programs. What you write is automatically saved to disk. This means you can work with unreliable power supplies, or learn by trying out new things and make what otherwise would be catastrophic mistakes.
You can experiment and learn.
With Emacspeak, you can listen to your email, browse the Web, write reports, programs, novels, and poetry.
You may also want to download various books, documents, and audio recordings from the Gutenberg Project archive, such as Jane Austen's novel Pride and Prejudice.
You can visit the archive directory using W3 mode or W3M mode at
ftp://metalab.unc.edu/pub/docs/books/gutenberg/
I just typed
Meta-x w3m-goto-url
and then the URL to reach that directory. (Well, actually, I typed
Meta-x w3m-got
and then pressed return. Emacs automatically completed the name of the command, so I did not have to do so much typing. And I copied the long URL listed above into the minibuffer so I did not have to type it. I hate typing and avoid it as much as possible.)
The index is
GUTINDEX.ALL
and you can download it using W3M mode by positioning point on the name and pressing the return key. (There are other ways of downloading it, too; this is one of them.)
As I write this in March 2006, the Gutenberg Project index is 2 megabytes long and lists more than ten thousand books, documents, and audio recordings. I saved my copy to a file, since I dislike long downloads.
You may want to use Emacspeak to listen to Pride and Prejudice, Alice in Wonderland, The Federalist Papers or other of the Gutenberg books, documents, and music.
The text files from the Gutenberg Project that I have downloaded use the DOS `carriage-return linefeed' convention to mark ends of lines rather than the Unix `newline' convention. This does not really matter since Emacspeak automatically detects the convention and visits the file using the DOS end-of-line coding format.
However, I prefer the Unix end-of-line format. After visiting a file, I run the set-buffer-file-coding-system command on it:
Control-x RET f unix RET
and then save the file:
Control-x Control-s
Here is a sample ~/.emacs file to customize Emacspeak. As far as I can figure, the only person who likes Emacs in its default configuration is Richard Stallman, who wrote it. All the rest of us change it somewhat.
I will not describe my full ~/.emacs file since it is more than 180 kilobytes long. Instead, I will create for you a much shorter, sample ~/.emacs file that works — I tested it – and that has a few of the customizations that I like.
The following ~/.emacs file contains left and right angle brackets. To hear them in Emacspeak, you must set the punctuation mode to all with the command:
Control-e d p all RET
In the ~/.emacs file itself, I have set the global default mode to none. To set a global default value of all in your ~/.emacs file, write the expression as:
;; The `t' is optional and sets the global default value.
;; Without it, the setting applied only to the local buffer.
(dtk-set-punctuations 'all t)
Here is the file:
;;; ~/.emacs -*- mode: emacs-lisp -*-
;; Robert J. Chassell, 2003 Apr 23, 2005 May 17
;;; First, general customizations for GNU Emacs
;; Make Text mode the default, with autofill
(setq default-major-mode 'text-mode)
(add-hook 'text-mode-hook 'turn-on-auto-fill)
;; Put two spaces after a colon with explicit fill commands
(setq colon-double-space t)
;; Do not insert tabs
; Indentation can insert tabs if indent-tabs-mode is non-nil,
; but I do not like tabs.
(setq-default indent-tabs-mode nil)
;; Insert newline at the end of a buffer
(setq next-line-add-newlines t)
;; Find existing buffer, even with different name
; (Avoid problems with symbolic links.)
(setq find-file-existing-other-name t)
;; Set command history list to 1000; it is 32 by default.
(setq list-command-history-max 1000)
;; Enable numbered backups
; `t' Make numbered backups.
; `nil' Make numbered backups for files
; that have numbered backups already.
; Otherwise, make single backups.
; `never' Always make single backups.
(setq version-control t)
;; By default the following are disabled to prevent confusing novices
;; Enable goal column
(put 'set-goal-column 'disabled nil)
;; Enable upcase region
(put 'upcase-region 'disabled nil)
;; Enable downcase region
(put 'downcase-region 'disabled nil)
;; Enable narrowing
(put 'narrow-to-region 'disabled nil)
;; Allow narrowing to a page.
(put 'narrow-to-page 'disabled nil)
;; Run Ediff with its help window in the same frame
(setq ediff-window-setup-function 'ediff-setup-windows-plain)
;; Ignore case when using `grep'
; The options are:
; -i Ignore case distinctions
; -n Prefix each line of output with line number
; -e Protect patterns beginning with -.
(setq grep-command "grep -i -n -e ")
;; Reduce frequent garbage collections
(setq gc-cons-threshold 600000)
;; Increase undo limit
(setq undo-strong-limit 60000)
;; Increase `max-specpdl-size'
(setq max-specpdl-size 2500)
;; Load Info at start up rather than when you first use Info
(load "info")
;;; Custom keybindings
;; Binding for goto-line.
(global-set-key "\C-c\C-g" 'goto-line)
;; Use `apropos' instead of `command-apropos'
(global-set-key "\C-ha" 'apropos)
;; Binding for compile
(global-set-key "\C-xc" 'compile)
;; Binding for occur
(global-set-key "\C-co" 'occur)
;; List all the function definitions and defvars and the like
;; that are in an Emacs Lisp source file.
(defun occurrences-of-def ()
"Run `occur' to find `^(def' from beginning of buffer or narrowed part.
Replaces going to the beginning of the buffer and finding the
occurrences of `^(def'."
(interactive)
(let ((buffer (current-buffer))
(current-def nil))
(save-excursion
(end-of-line) ; so as to find def on current line
(if (re-search-backward "^(def\\w+\\W+\\w+-*.*" nil t)
(setq current-def
(buffer-substring (match-beginning 0) (match-end 0)))))
(save-excursion
(goto-char (point-min))
(occur "^(def\\w+-?\\w*" nil)
)
(pop-to-buffer "*Occur*")
(goto-char (point-min))
(if current-def (search-forward current-def))
(beginning-of-line)
))
(global-set-key "\C-cf" 'occurrences-of-def)
;; For sighted people using GNU Emacs 21, turn off the blinking cursor!
(if (fboundp 'blink-cursor-mode) (blink-cursor-mode -1))
;; Set the title for a frame
(setq frame-title-format '("Emacspeak: %b"))
;;; For Emacspeak specifically
;; Make sure all Emacspeak code is in the load-path.
;; I commented this out since the load-path since is
;; specific to my set up.
;; The rest of this ~/.emacs file should work with
;; all instances of Emacspeak.
;; Depending on how you set up Emacspeak, you may not need to
;; configure your load-path manually at all.
;; (setq load-path (cons "/usr/local/src/emacspeak/lisp/" load-path))
;; A more modern and elegant way to set your load-path is
;; (add-to-list 'load-path "/usr/local/share/emacs/site-lisp")
;; The older method reminds you what `cons' does; the newer is simpler.
;; Turn on global-font-lock for the voice lock engine
(load "font-lock")
(global-font-lock-mode 1)
(setq font-lock-global-modes t)
;; (setq global-voice-lock-mode t) ; not needed and obsolete in Emacs 21
;; Speak time in a reasonable format: `C-e t'
(setq emacspeak-speak-time-format-string
"The time is %_H hours %M minutes %Z on %A %_e %B %Y")
;; Set the audio theme
;; /usr/local/src/emacspeak/sounds/chimes-mono/
;; ( alternatively /usr/local/src/emacspeak/sounds/chimes-stereo/ )
(emacspeak-sounds-define-theme
(expand-file-name "chimes-mono/" emacspeak-sounds-directory)
".wav")
;; Set punctuation mode to MODE `\'some, 'all, or 'none
;; For individual buffers the keybinding is: `C-e d p'
;; The expression in this ~/.emacs file sets the mode globally.
;; In general, I prefer 'none but then you will not hear
;; braces and brackets spoken. To listen to this ~/.emacs
;; file, you need to set the mode to 'all.
;; That is best done using the `C-e d p' keybinding.
;; With an interactive prefix argument, `C-u C-e d p',
;; set the global default value.
;; In this sample, I have set the mode to 'none.
(dtk-set-punctuations 'none t)
;;;;;;;;;;;;;;;; end ~/.emacs ::::::::::::::::
At the time I write this in May 2005, I am running two instances of
Emacspeak, one using the eflite software text-to-speech
synthesizer and the other using the espeakf software
text-to-speech synthesizer.
The current espeakf provides different voices, which my current
version of eflite lacks. Also, in my set up, it sometimes
speaks pronunciation when it should not; but not always.
I prefer espeakf.
At the moment, I am using two different Emacspeak commands, one for
starting Emacspeak with eflite and the other for starting Emacspeak
with espeakf. Right now, I am running both, each as its own instance
of Emacspeak.
The two starting commands are:
emacspeak-eflite
emacspeak-espeak
The two commands are shell scripts in the /usr/local/bin/ directory.
The regular Emacspeak command is in the /usr/bin/ directory, but since these two commands are experimental, I am putting them into the /usr/local/bin/ directory, to remind me that I created them locally.
(Incidentally, I am also running two other non-Emacspeak instances of Emacs: an experimental Emacs version 22 owned by user `bob', and an Emacs version 20.7 owned by user `root'. That way, I separate dangerous system administration task from ordinary work.)
espeakfRegarding installation: if you have not installed Emacspeak recently, and plan to install the package again soon for someone else, please practice a reinstall first. Perhaps, nowadays, all will go well; but perhaps not.
In the past, I have had troubles with Emacspeak installation, although in 2003 and more recently, everything has gone well. I am sighted, and do not depend on Emacspeak, so perhaps I continually renew my ignorance.
In addition to a computer with a sound system, Emacspeak requires three packages:
A hardware text-to-speech synthesizer is best, but I do not have one. Instead I use two different software text-to-speech synthesizers, both of which are immature programs that make use of the more basic Festival text-to-speech program.
Installation is extensively described in the Emacspeak manual. See Installation.
Also, James R. Van Zandt has written an installation HOWTO that is now maintained by Sharon Snider. It is in the Emacspeak sources and on the Web at http://www.tldp.org/HOWTO/Emacspeak-HOWTO/.
Please consider building and installing your software from source. This process is straightforward and I have done this in the past with great success.
However, I am going to describe only what you do to install
espeakf, which is not in Debian, and what you do in Debian to
install eflite. The process should be simple, but for some
reason, possibly because I tend to use unstable, experimental
software, I have not always found it so.
In Debian, as of April 2003, to install Emacspeak using the eflite software text-to-speech synthesizer, type the following in a shell:
apt-get install emacs21 emacs21-el emacspeak eflite
The emacs21-el package contains the Emacs Lisp source files that Emacs uses. In theory, you can run Emacs without them, but I have found them essential, if only for the extra documentation and commentary they provide. It does not matter if some of the packages are already installed or will be installed automatically.
After running apt-get, you should be able to run Emacspeak by typing
emacspeak
in a shell. That command is a shell script in /usr/bin/emacspeak.
In theory, that is all there is to it. And if you are not using the awkward, non-standard system that I am, but instead are using a well configured, modern, standard system, what I just told you should work.
Since I run both the eflite and espeakf text-to-speech
synthesizers, and at this stage am running them in what amounts to a
local experiment, I copied the shell script that starts Emacspeak with
eflite from the /usr/bin/ directory to the /usr/local/bin/ directory
and renamed it to /usr/local/bin/emacspeak-eflite.
I start Emacspeak with espeakf with a different shell script,
called /usr/local/bin/emacspeak-epeakf. (These names are not
very imaginative, but I can remember them.)
espeakf is available from CVS on SourceForge. Fortunately, it
is easy to download and install.
First you need to decide on a directory in which you plan to keep the
espeakf directory that you will download. I use
/usr/local/src/
Make that directory if you do not have it already, then go to it.
Here are the commands, which I run as the non-privileged user `bob' in a shell, such as one in Emacspeak that you can start with the Meta-x term command.
This facility is described in
(info "(emacspeak)Running Terminal Based Applications")
(Before you try Meta-x term, be sure to learn how to get out of
the buffer! In Term mode, the usual commands for killing a buffer or
switching buffers do not work. I find Control-c Control-f, the
new keybinding for find-file useful, as well as Control-c
k, which kills the buffer.)
In a shell, I first switch to the appropriate directory, and then I run the CVS login command:
cd /usr/local/src/
cvs -d :pserver:anonymous@cvs.espeakf.sourceforge.net:/cvsroot/espeakf login
You need to login once for CVS. The login command will ask for your password. Just press your return key. Then, type the following command:
cvs -d :pserver:anonymous@cvs.espeakf.sourceforge.net:/cvsroot/espeakf co espeakf
These commands will create a /usr/local/src/espeakf/ directory
below your /usr/local/src/ directory and fill it with several
files and other directories. To make use of the espeakf
text-to-speech synthesizer, Emacspeak runs an executable Perl script
called espeakf.pl. CVS installs that Perl script properly, so
you do not need to do anything.
However, you do need to modify your /usr/local/bin/emacspeak-eflite and /usr/local/bin/emacspeak-espeakf start up scripts for the two different text-to-speech synthesizers.
I use a variation on the /usr/bin/emacspeak script that
Emacspeak creates automatically. That script uses information from a
/etc/emacspeak.conf file. I ignore that configuration file
because the configuration for Emacspeak with the two synthesizers is
different. Instead, I put all the configuration information into the
two /usr/local/bin/ scripts.
The two scripts contain sharp signs, exclamation marks, left and right brackets, and much else. To listen to all the characters in them, you need to set Emacspeak's punctuation mode to all. If you are listening to this in Emacspeak, you can evaluate the following expression by placing point after it and typing Control-x Control-e:
(dtk-set-punctuations 'all)
First, the /usr/local/bin/emacspeak-eflite script.
#!/bin/sh
## /usr/local/bin/emacspeak-eflite - execute emacs with speech enhancements
## use EFLITE
## based on
#$Id$
# with minor changes, 2006 Mar 18
if [ -f $HOME/.emacs ]
then
INITSTR="-l $HOME/.emacs"
fi
CL_ALL=""
for CL in $* ; do
if [ "$CL" = "-o" ]; then
DTK_PROGRAM=outloud
export DTK_PROGRAM
elif [ "$CL" = "-d" ]; then
DTK_PROGRAM=dtk-soft
export DTK_PROGRAM
elif [ "$CL" = "-q" ]; then
INITSTR=""
else
CL_ALL="$CL_ALL $CL"
fi
done
export EMACS_UNIBYTE=1
export DTK_PROGRAM=/usr/local/bin/eflite
export DTK_TCL=/usr/local/bin/eflite
exec emacs -q \
-l /usr/share/emacs/site-lisp/emacspeak/lisp/emacspeak-setup.el \
$INITSTR $CL_ALL
### end /usr/local/bin/emacspeak-eflite
Second, the /usr/local/bin/emacspeak-espeakf script:
#!/bin/sh
## /usr/local/bin/emacspeak-espeakf - execute emacs with speech enhancements
## use ESPEAKF
## based on
#$Id$
# with minor changes, 2006 Mar 18
if [ -f $HOME/.emacs ]
then
INITSTR="-l $HOME/.emacs"
fi
CL_ALL=""
for CL in $* ; do
if [ "$CL" = "-o" ]; then
DTK_PROGRAM=/usr/local/src/espeakf/espeakf.pl
export DTK_PROGRAM
elif [ "$CL" = "-d" ]; then
DTK_PROGRAM=dtk-soft
export DTK_PROGRAM
elif [ "$CL" = "-q" ]; then
INITSTR=""
else
CL_ALL="$CL_ALL $CL"
fi
done
export EMACS_UNIBYTE=1
export DTK_PROGRAM=/usr/local/src/espeakf/espeakf.pl
export DTK_TCL=/usr/bin/perl
exec emacs -q \
-l /usr/share/emacs/site-lisp/emacspeak/lisp/emacspeak-setup.el \
$INITSTR $CL_ALL
### end /usr/local/bin/emacspeak-espeakf
Both these files need to be made executable. I executed
chmod 755 emacspeak-*
in a shell as user `root'.
### End of document