Subject: [ecasound] not sure if I reported these
From: Jeremy Hall (jhall_AT_uu.net)
Date: Mon Nov 19 2001 - 16:48:32 EET
Hi,
not sure if I reported these but if I did please ignore. While working
with ecasound, I found some annoying features.
-eca: a feature causes the compressor not to kick in the first time it
needs to, causing the resulting volume to be quite a bit louder until the
compressor kicks in for the first time (like if you use -eca then -ea to
normalize)
-enm: gate won't open quickly enough, even when told to do so with 1ms,
resulting in syllables clipped upon gate open
-enm: threshold cannot properly determine between the wanted signal (gate
open) and unwanted signal (gate closed)
when muting loop devices, , the signal from their parents can still
sometimes be audible
when muting the first chane in a loop, spurious noises followed by process
halting but not saying it halts (using freeverb on the loop) a work around
is to -ea:0 that channel instead :(((
if multiple parent chanes of a loop are muted, resulting in only one left,
an artificial ringing is introduced that is aligned with
frames_per_cycle. the solution is to -ea:0 at least one of those muted
chanes and unmute it
when multiple chanes with multiple effects are active, a sudden explosion
in needing to compress or act on such effects may cause spurious xruns
even tho the cpu is not thoroughly busy
on a SMP system ecasound seems to only use one cpu
in some cases, (might be related to using f32_le wav files) -t: behaves
differently for output to alsa device and output to raw file
outputting to wav file when using 48khz may result in unwanted sample
conversions solution is to write to raw file then use sox to write the wav
file
converting from f32_le/s32_le to s16_le causes audible dither solution not
known. recommendations to do this greatly welcome
using noninterleaving and alsa with s32_le may result in LOUD
NOISE! (-f:s32_le,28,48000,n)
-f: sometimes auto-resets itself to something,2,44100 therefore
explicitely using -f: is necessary
when writing out a chane setup for the first time, the parser may write
s16_le,2,44100 regardless of the real fmt, channels, or rate. IT then
becomes necessary to go in and manually change it. It usually gets the
channel count right at least, but if a wav file is f32_le, you need to go
explicitely set that in the freshly-written file.
using the -y: option prevents you from using fw, rw, or setpos in the
session because using these will cause all your audio to desync itself. We
need a better solution than using aio-rw etc for each io object.
time is not accurate enough when written to a chane setup file and may
result in desync because you cannot say go to frame X in a source.
the -ec: compressor works great as an overdriver.
-eli:1073 only accepts values with hundredths precision, .003 turns into
0.0
despite the discussion in the manpage, the -eca ``fastrate'' and ``rate''
are not well understood
in the appended setup file, for some reason chane 17 is not panned
properly--it drifts to the right, along with 6.
Peace:/tape1/pr-20011108-session/sounds/eca # cat prb.fil
# ecasound chainsetup file
# general
-b:64 -r:50 -z:intbuf -z:nodb -sr:48000 -n:prb -X -z:noxruns -z:psr
-t:1095
# audio inputs
-a:4 -f:f32_le,1,48000 -i:audio4.wav, -y:2.645
-a:6 -f:f32_le,1,48000 -i:audio6.wav, -y:2.893
-a:7 -f:f32_le,1,48000 -i:audio7.wav,
-a:8 -f:f32_le,1,48000 -i:audio8.wav,
-a:11 -f:f32_le,1,48000 -i:audio11.wav, -y:2084.139
-a:12 -f:f32_le,1,48000 -i:audio12.wav, -y:2084.139
-a:13 -f:f32_le,1,48000 -i:audio13.wav, -y:2084.139
-a:15 -f:f32_le,1,48000 -i:audio15.wav, -y:2084.139
-a:16 -f:f32_le,1,48000 -i:audio16.wav, -y:2084.139
-a:17 -f:f32_le,1,48000 -i:audio17.wav, -y:2084.139
-a:21 -f:f32_le,1,48000 -i:audio21.wav, -y:2084.139
-a:22 -f:s32_le,2,48000 -i:loop,1,1
-a:23 -f:s32_le,2,48000 -i:loop,2,2
-a:25 -f:s32_le,2,48000 -i:loop,4,4
-a:26,27 -f:s32_le,2,48000 -i:loop,5,5
# audio outputs
-a:4,6,17 -f:s32_le,2,48000 -o:loop,1,1
-a:11,12,13,15,16,21 -f:s32_le,2,48000 -o:loop,2,2
-a:7,8 -f:s32_le,2,48000 -o:loop,4,4
-a:22,23,25 -f:s32_le,2,48000 -o:loop,5,5
-a:26 -f:s32_le,2,48000 -o:alsa,dsp1
-a:27 -f:s16_le,2,48000 -o:/tmp/out.raw,
# chain operators and controllers
-a:4 -erc:1.00,2.00 -eli:1073,0.09,0.70,0.03,0.50 -epp:30.00 -ea:85.00
-a:6 -erc:1.00,2.00 -eli:1073,0.09,0.70,0.03,0.50 -epp:70.00 -ea:40.00
-a:7 -epp:0.00
-a:8 -erc:1.00,2.00 -epp:100.00
-a:11 -epp:0.00 -ea:159.00
-a:12 -erc:1.00,2.00 -epp:100.00 -ea:159.00
-a:13 -erc:1.00,2.00 -epp:25.00 -ea:350.00
-a:15 -erc:1.00,2.00 -epp:50.00 -ea:130.00
-a:16 -erc:1.00,2.00 -epp:75.00 -ea:16.00
-a:17 -erc:1.00,2.00 -epp:50.00 -eli:1203,2.00,600.00,2.00 -ea:150.00
-enm:2.00,340.00,30.00,1.00,1.00 -eli:1073,0.10,0.60,0.05,0.50 -ea:60.00
-a:21 -erc:1.00,2.00 -epp:90.00 -ea:76.10 -ea:100.00
-a:22 -eli:1123,0.00,0.85,0.30,0.05,0.50,1.00 -ea:100.00
-a:23 -ea:200.00
-a:25 -ea:43.00 -eli:1203,2.00,130.00,1.50 -eli:1203,3.00,1500.00,6.00
-eli:1123,0.00,0.75,0.70,0.13,0.50,1.00
-a:26 -ea:700.00
-a:27 -ea:700.00
Peace:/tape1/pr-20011108-session/sounds/eca #
-- To unsubscribe send message 'unsubscribe' in the body of the message to <ecasound-list-request_AT_wakkanet.fi>.
This archive was generated by hypermail 2b28 : Mon Nov 19 2001 - 16:42:23 EET