Thomas Stivers <stivers_t at tomass.dyndns.org> writes: > Hi all. There are several small things I don't quite understand and I > would like to explain them in my speakup howto (you know the one I have > been talking about for some time but still been to lazy to finish). > > First there are several items in /proc/speakup that I don't really get. > These are cursor_time, delay_time, ex_num, full_time,jiffy_delta, punct, > say_control, say_word_ctl, and trigger_time. Some explanation on some or Let's see if I can help with a few of these: cursor_time - the amount of time in milliseconds from the time the cursor moves to the speaking of the line/character/whatever. A value of 120 means that on a down arrow the line will be spoken 120 ms after the key has been hit. puncht - Chuck already pointed out is the hardware synths punctuation level which we set so we can provide our own version of punctuation and number processing and the like. This next group of four are related to synthesizer and speakup control times and are critical for smooth or choppy behaviour. full_time - the amount of time in milliseconds to sleep when a synth announces it's buffer is full before trying to send the synth more data. trigger_time - is the time in milliseconds between when the last character was received and when we should send buffered data to the synth. If it is to large you will get sluggish responsiveness when typing or partial line prompts show up and the like. If this is to low then you will hear characters and partial words instead of complete words and lines on output from the computer. delay_time - is the time in milliseconds between speakup sending data to the synth. So it sends data and then goes to sleep for delay_time ms. If this is to large you will get choppy output because the synth will finish before new material is sent to it. If it is to low once again you will get partial words and lines because not enough new material has come in before speaking. jiffy_delta - the amount of time we allow ourselves to hold the processor when sending data to the synth. When we have the processor nobody else can have it so if we hold it to long then all processes get delayed and the whole system becomes unresponsive. Tuning the above values to maintain good responsiveness and smooth speech output can be very easy with some synths and a real pain with others. The dectalk family for instance has some very long command strings so if you don't have the values set correctly you only get partial command seequences sent out before we go to sleep and the synth times out itself. The other controls I will have to look up myself because I don't remember what they all do off hand myself. Kirk -- Kirk Reiser The Computer Braille Facility e-mail: kirk at braille.uwo.ca University of Western Ontario phone: (519) 661-3061
