Subject: [ecasound] found some bugs in latest development versions
From: Jeremy Hall (jhall@UU.NET)
Date: Mon Apr 24 2000 - 10:52:05 EEST
well yesterday I attempted to use ecasound for some local mixing, and came
up with some startling concerns.
1: if one accidentally specifies a -i argument, but argument is not a
valid file for some reason, ecasound dumps core.
2: if one uses -kl on a nonexistant argument, ecasound dumps core, for
ecasound -i /tmp/in.raw -o /tmp/out.raw -c
in this example, we forgot -ea:100
3: it is still pretty difficult to use ecasound practically in some
circumstances, like if one of the inputs isn't providing input, ecasound
4: it would be nice if we could have a "audition" mode, one way to do this
would be to use two versions of ecasound, with two inputs, one being te
"live" signal and another being a monitor signal. the output is the
/dev/dsp, or the alsa equivilent.
5: I am not certain how to make ecasound add a new input gracefully, then
manually start this input. I don't want to continuously add new channels,
but I do want to be able to switch the input to a new file.
6: have the ability to enable a "last-resort" input. This would be if
something goes wrong, and an input source just stops for some reason. If
this is the only input, SOMETHING needs to be transmitted in a live
situation, maybe fire off a spot or some sort of alert signal that other
computers could detect and do something useful with.
maybe what I need to do is write a separate application for what I am
trying to do. What AM I trying to do you might ask?
well, I've got tons of files, cds, and other input devices that people
have given me, then they say, "ok, make us some live music for X hours"
As you know, this isn't always the easiest job, and sometimes one finds it
difficult to do all this typing. For network use, they want timed
events. (like news insert on top of the hour) they want me to respond to
25HZ tones, for example. They want NO dead air, and if it DOES appear,
currently it requires lots of typing and many commands to restore
In short, this fun project to provide music turned into a desaster,
because of the complexity required to do something. for example, to do a
simple action like pot down a channel gracefully you have to type
now in seven seconds, channel1 will be muted. Perhaps I have missed some
new feature that makes that a little more exciting.
now let's say that 1 is playing a song that will end in 5 seconds. I need
to queue the next song on 2 so that it is ready for when I want to start
2, not a predetermined event (like ewf)
I didn't have time to write down other fun realizations, I should
have. perhaps the next time I do this, I will have more time to come up
with good solutions.
oh, almost forgot,
and for some reason, ecasound didn't play an mp3 file all the way
through. NO idea what that's all about, but I had some mp3 files that were
all recorded at the same bitrate, and because I couldn't tell ecasound to
just start playing all these files in a series as fasst as possible
(within reason), I just catted all of them into a large file. ecasound
played 5 of the songs, then stopped at the end of one of them, where one
mp3 might stop and the next might start. No matter what I tried, ecasound
would NOT go past that point, even if I told it to fastforward etc. This
created some fast thinking problems, ultimately resulting in lots of dead
air as I tried to figure out why ecasound wasn't responsive. as I typed
setpos 0, cs-connect, t, nothing happened. Sometimes, it would take on the
order of tens of seconds before ecasound would continue playing SOMETHING,
ANYTHING to keep the flow moving.
-- To unsubscribe send message 'unsubscribe' in the body of the message to <email@example.com>.
This archive was generated by hypermail 2b28 : Mon Apr 24 2000 - 10:53:04 EEST