Re: [ARSCLIST] Heavily distorted signal

[David S Sager <dsag@xxxxxxx>]

Diamond Cut has several different forensics filters available in their
millenium and DC5 editions.  I have used neither.

DS

>>> trescott@xxxxxxxxx 04/17/03 01:46AM >>>
I've lost the original post, but has anyone used audio forensic
software?
Here is one outfit and there is at least one other in Europe:
http://www.intdevices.com/

Paul T. Jackson - Trescott Research
Information Resources and Library Development
trescott@xxxxxxxxx <mailto:trescott@xxxxxxxxx>
http://www.bookbay.com/PioneersInBrass.htm


-----Original Message-----
From: Association for Recorded Sound Discussion List
[mailto:ARSCLIST@xxxxxxx]On Behalf Of Mike Richter
Sent: Wednesday, April 16, 2003 11:50 AM
Subject:


Subject:      Re: [ARSCLIST] Heavily distorted signal
Comments: To: Association for Recorded Sound Discussion List
<ARSCLIST@xxxxxxx>
To: ARSCLIST@xxxxxxxxxxxx
In-Reply-To:  <EB14F3BB8D31184D902B2B6209A1F87207182B35@arion>
X-Server: High Performance Mail Server - http://surgemail.com
X-Smite-CRC:
A#1nGKQX6$1a8yLaQ$n1yIYv#HNsbFH#kZcu02#1yIj62W#1CA6gmw#1hE5y6s$1JS693y

At 11:36 AM 4/16/2003 -0500, Peoples, Curtis wrote:
>I have a copy of some Turkish folk poets speaking with and without
>music. The original analog tape is very distorted. Are there any
tricks
>or techniques to make the voice sound clearer. Is there a certain
>frequency range I can work with for distortion? Are there any
software
>programs or outboard gear to remedy this problem? Any advice is
>appreciated.

For some reason, two replies arrived at my in box before this original
message.

You should play with equalization (which may vary from speaker to
speaker)
to optimize retrieval. Frequently in live recording, subsonic and low
frequencies are excessive and overload the electronics. You cannot
correct
for the recording overload, but you can reduce the playback effects by
judicious filtering. Experiment for your material; I often use a
fourth-order filter at about 150 Hz.

The highs may be loaded with distortion further reducing
intelligibility
but the formants are important. For a male speaking voice, try about a
6 db
boost from 1-2 KHz followed by a 3 db per octave rolloff. You may want
to
move it up a half octave for female voices, but again the best
intelligibility will depend on the faults of the original and the
timbre of
the speaker.

Most denoising (I've not used the Sonic Foundry plugins) will be
ineffective at low settings and objectionable for artifacts (metallic
sound, audible echo) when set high enough to be useful. Analogue
autocorrelation has proved more effective for me than digital, so I
occasionally pull out an ancient Phase Linear autocorrelator for
really
noisy sources.

If you have a short sample for me to try digitally, please contact me
off-list.


Mike
mrichter@xxxxxxx
http://www.mrichter.com/