Lynn,
I was not able to DL your zip File, so I don't know about the details of
your problems. However, referring to the dynamic range Compressor
Effect, are you aware of Build 1.2.4?
So far I've used build 1.2.4. to Compress about 20 hours of Classical
music. I am _very_ satisfied with the results, and plan on much more
continued use of it.
This includes compressing as much as a 50dB dynamic range to a 20 dB DR
(using: Threshold 50 dB, Slope 2.5:1, and Attack Time 0.1 sec).
=> Very Importantly: Uncheck "Apply gain after compressing"
This secondary function operates very badly:
It produces severe clipping on crescendos.
=> Instead use the Amplify effect, and be sure to check
"Don't Allow Clipping"
(I then may reduce the level by up to 3 dB.)
There is a minor imperfection in applying Compress as above:
The actual attack time is not zero, so on a very abrupt crescendo, a
brief overshoot may occur. This only occurs rarely with actual music
material, but when it does it's generally about 1 to 3 dB (although I've
seen as much as 6 dB).
One can get a larger overshoot effect using a test tone:
For example, I used a 400 Hz triangle wave that has an abrupt 24 dB
step, going from -24 dB to 0 dB from one cycle to the next. I used
Compress, with a 2:1 slope, and Amplify as above. This resulted in
a 12 dB overshoot that settled out in about 40 msec.
There is a similar problem on a large, sudden (less than 500 msec)
decrease in level: undershoot can occur. Although I've rarely (and
barely) noticed it with actual music, this can be readily observed using
a abrupt step decrease in a test tone. (The recovery time is dependent
on the step dB decease, and can be up to 400 msec.)
=== === === ===
Let me emphasize that the Compressor Effect in 1.2.4 works _far better_
than that of other products that I have tested, including commercial.
AndyB
Lynn Allan wrote:
> * Replies will be sent through Spamex to
audacity-devel@...
> * For additional info click ->
http://www.spamex.com/i/?v=6850724>
>
>>Leveler: What is the reason to use the leveller rather than the
>
> compressor?
>
> Summary:
> problems with the compressor effect
>
> Details:
> <alert comment="not an audio engineer">
>
> Earlier in the year, I had heard that there were flaws in the
> Compressor effect, which I verified with some sine wave tests (see
> attachment with screen shots ... spikes and clipping at the start).
>
> This problem with the Compressor effect seemed to be low priority, and
> it didn't seem like any progress was being made on a fix. Therefore, I
> implemented the simplified "Leveler" effect, which was initially to be
> strictly for CleanSpeech. At that time, CleanSpeech was a "spin-off"
> of Audacity and specialized for speech processing (rather than being
> more of a "mode" like it is now). "Leveling" is perhaps the most
> important effect that CleanSpeech provides, so I considered it high
> priority.
>
> I am not at all an audio engineer, but the Leveler effect has
> something of a "banana shaped" amplification curve with six adjustment
> bands. There is an end-user specified "inflection point" below which
> signals are rolled off. Above this inflection point (e.g. -55db), the
> signal is amplified for a range, and then rolled off to cut back on
> the loudest signals. This is a relatively fast effect since no FFT is
> involved, so "lighter" and "heaver" leveling is accomplished by making
> multiple passes to "smooth" the transitions between the bands.
>
> I used two test files:
> 1hz sine wave at 0db at 192,000 sampling rate to be able to see what
> was going on
>
> 50hz sine wave at 0db at 22050 sampling rate to be able to hear what
> was going on
>
> Here is a link to these files, and before/after screen shots of using
> the Compressor effect at its default settings.
>
>
http://cleanspeech.sf.net/misc/Leveler_Compressor_Tests.zip>
> </alert>
>
>
>
>
> -------------------------------------------------------
> SF.Net email is Sponsored by the Better Software Conference & EXPO
> September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
> Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
> Security * Process Improvement & Measurement *
http://www.sqe.com/bsce5sf> _______________________________________________
> Audacity-devel mailing list
>
Audacity-devel@...
>
https://lists.sourceforge.net/lists/listinfo/audacity-devel>
>
-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement *
http://www.sqe.com/bsce5sf_______________________________________________
Audacity-devel mailing list
Audacity-devel@...
https://lists.sourceforge.net/lists/listinfo/audacity-devel
