I think these ideas are some of the best I've heard so far.
Especially the auto-normalize. That seems so useful and
after all these years, I never though of that one and no one
else has suggested that.


On 4 Aug 2005 23:05:31 -0700, wrote:

I've been using your program for some time. Here are my suggestions:

1) Multichannel with individual squelch would sure work for me, and
quite frankly it would be the only way to get me to pay for the
software. I'm sure there is a pro-audio multitrack file format.

2) Some sort of periodic auto-save function is needed. I often record
in the field and don't always catch the notebook battery meter
indicating a low function.

Would auto-save to a new file name (say, using a sequence
number) be acceptable instead of re-opening the previously
written file?

A .wav or .mp3 file has to be closed before it's considered
to be safely stored. That means that it cannot be appended
to without a lot of potential problems. If the format is raw
data with no headers (like the .RAW format used by Sound
forge) then it would be easy to append after closing the
file. Maybe I should consider adding a .RAW format for that
reason.

But IMHO, it's not safe to update the headers (which is
necessary to append more audio) of the .WAV or .MP3 file
that's already been closed because the computer could die in
the middle of the update and that might leave the file
crippled. Manipulating the previously saved headers can
invite other problems too, such as adding a lot of disk
seeking and CPU time to reopen the file for appending after
it's already been closed. Appending more audio requires a
read-scan of the entire file in order to locate and change
the existing audio (RIFF) headers.

Note that I always record in PCM and then convert to compressed audio
later. This has a few advantages. One, I can "normalize" the audio in
soundforge prior to compressing it. If you adjust the audio level of a
compressed file, I believe it gets compressed twice. Second, I I can
take individual bits of audio and boost the level of just that section.
Audio levels are all over the map if you record military air. Third,
the filters work better on PCM. Often I need to notch 400hz and 1200Hz
from aircraft generators.

Now this leads to another featu

3) Dump the auto-clip scheme.

If you're talking about the Scanner Recorder anti-clip, that
has to do with preventing the audio from being clipped off
at the beginning the way a mechanical VOX recorder does.
That works by keeping the last 10 MS of audio in a buffer
and drawing from that buffer whenever the VOX is opened to
prevent that pop sound. Like if it records a person saying
the word "testing" and it ends up sounding like "esting".

Sixteen bits of audio is plenty of
headroom. Encourage the user to leave a bit of headroom when recording.
Then for each segment of audio, normalize it. This will adjust for
variations in audio level from radio to radio. This implies you will
need a buffer for the audio segment, and the ability to simultaneously
normalize one segment of audio while still recording the next segment.

Here is an idea for another product. These flash recorders (Iriver,
Creative, etc) are getting quite cheap, but they still haven't mastered
the idea of a vox. A program that would take a file from a flash
recorder and strip out the silence would be useful.

Sound Forge can do that using Auto-Region I think.

Ultimately, the audio recorder belong in the scanner itself. The
manufacturers are idiots for not doing so. It would have the advantage
that the audio segment could be tagged with frequency, time, talk
group,etc.You could scan multiple frequencies and then reassemble all
the audio segments using the tags. Uniden and GRE just can't think out
of the box. For P25, you could just store the digital audio.

I really appreciate the great ideas! Thanks.