the direction of speakup

Brian Buhrow buhrow at
Fri May 10 13:27:15 EDT 2013

	Hello.  If you're interested, I have a bunch of patches that make Yasr
much more stable.  When I worked on it, I found the problem was that the
communication between Yasr and the speech engine (eflite in my case) was
very unstable.  Eflite would take a dump and Yasr would continue writing to
it as if nothing was wrong.  That still happens, but my patches detect when
the speech engine has disappeared and respawn it and reconnect output to
the new speech instance.  Without these fixes, I would agree that Yasr was
pretty unusable.  I submitted the patches back to the original author for
Yasr, but I don't know if he incorporated them into the official

On May 10,  1:37pm, Mike Ray wrote:
} Subject: Re: the direction of speakup
} Kyle,
} It was totally unstable on Raspberry Pi.  I tried rebuilding it and 
} speech-dispatcher.  I tried it with espeak and festival and it was only 
} able to run for a few minutes before crashing.
} Now I have the UART cable I can debug it as well.  Maybe it's the same 
} thing that's crashing SpeakUp, in which case it's probably something 
} about the architecture or the Broadcom sound driver.
} yasr also hasn't seen any work for about 5 years.  I did speak to the 
} developer and he wants to pass it on to someone else.  So if I can get 
} my head round why it doesn't work on Raspberry Pi maybe I will have a go.
} Mike
} On 10/05/2013 13:30, Kyle wrote:
} > According to Jason White:
} > # It runs its own shell and captures input/output, somewhat like screen(1).
} >
} > This actually makes YASR the most portable text console screen reader I
} > am aware of, since it can run on just about any Unix-like operating
} > system. It runs entirely in userspace and depends on shell output rather
} > than relying on any kernel level code or output. It also has the benefit
} > of being able to work with a wide range of hardware synthesizers via
} > Emacspeak servers and possibly other local drivers as well, and also has
} > software speech available through various interfaces, including EFlite
} > and speech-dispatcher. The trade-off is that you will get no speech
} > prior to login, although with the correct login script, you can have
} > YASR come up automatically once you've logged into the console you want
} > to use. There once was a separate program included in the YASR source
} > tree that could read the console prior to login, but I don't currently
} > know if it still works. I remember getting it to work at one point, but
} > that was some time ago. I did most things with a single text console
} > that ran YASR automatically at login and did all my work in Screen,
} > which allowed me to have a nearly unlimited number of "windows" open on
} > a single console, all under a single YASR instance.
} >
} > Just a quick note: because of the way YASR works in a subshell, it
} > should be capable of working in a desktop terminal application like
} > Xterm, giving you familiar functionality when you have that text-based
} > application that Orca doesn't like in gnome-terminal. Keybindings
} > shouldn't be a problem either, even if you use gnome-terminal and
} > silence Orca, since as far as I know, there aren't any conflicts between
} > Orca and YASR keyboard commands.
} > ~Kyle
} >
} -- 
} Michael A. Ray
} Analyst/Programmer
} Witley, Surrey, South-east UK
} Interested in accessibility on the Raspberry Pi?
} Visit:
}  From where you can join our mailing list for visually-impaired Pi hackers
} _______________________________________________
} Speakup mailing list
} Speakup at
>-- End of excerpt from Mike Ray

More information about the Speakup mailing list