Bluenote ha scritto:Cosa fa esattamente la routine psicoacustica (PA) ?
Facendo una ricerca qui sul forum, ho letto solo che serve per evitare di intervenire con la target.
Ho fatto qualche prova ieri pomeriggio e ho avuto la sensazione ad orecchio che "aggiungesse" qualche armonica su alcune frequenze.
Fa tante di quelle cose che conviene postare l'intero paragrafo relativo alla PA
4.3 Psychoacoustic target computation
Starting from version 3.0.0 an optional stage, used to compute a target frequency response based on the psychoacoustic perception of the corrected frequency response, has been introduced. The target response is based on the computation of the magnitude response envelope of the corrected impulse response, which is an extension of the spectral envelope concept. This is performed before the application of the usual target response, so that the standard target response is not compensated back by this stage.
The spectral envelope is a concept which has been introduced in the field of speech synthesis and analysis and is defined simply as a smooth curve connecting or somewhat following the peaks of the signal spectrum. There are strong arguments and experimental evidence supporting this approach and the idea that our ear uses the spectral envelope for the recognition of sounds. The spectral envelope, for example, allow our ear to understand speech under many different conditions, whether it is voiced, whispered or generated by other means. These different conditions generate completely different spectrums but usually pretty similar spectral envelopes. The spectral envelope also easily explains why our ear is more sensitive to peak in the magnitude response and less sensitive to dips. A curve based on the peaks of the magnitude response is by definition little or not affected at all by dips in the frequency response.
In the speech recognition field many procedures have been developed to compute the spectral envelope. Some of them are based on Linear Predictive Coding (LPC), the Discrete Cepstrum, the so called “True Envelope” and finally the Minimum Variance Distortionless Response (MVDR). Most of these methods are optimized for speed, noise resilience and to provide good results in the voice spectrum range sampled at low sample rates, so they are not really suited for HiFi usage.
Within DRC a different procedure has been developed. This is a variation of the usual fractional octave smoothing procedure, using the parametric Hölder mean instead of the usual simple averaging. Furthermore the smoothing has been extended to provide the Bark and ERB scales resolution when applicable.
Figure 9: Example of a magnitude response envelope. The unsmoothed magnitude response of a typical room, corrected with a flat target, is plotted. Superimposed there is the usual smoothing, computed on the ERB scale, showing an essentially flat magnitude response, as expected. The magnitude response envelope, computed on the ERB scale too using standard parameters, show a rising slope which is in good agreement with the inverse of one of the target responses suggested in the literature.
For the tuning of the parameters used in the magnitude response envelope computation some typical real world room magnitude responses have been taken. The computation parameters have been set so that the resulting magnitude response envelope provides a target response as close as possible to the inverse of the usual target responses suggested in the literature. An example is reported in picture 9. The same parameters have been tested also in some not so common room to check that they were still providing the expected results. Of course, considering that now the basic target response is provided by the inverse of the magnitude response envelope, the usual target responses are no longer needed, apart from subsonic or ultrasonic filtering, and should be set to flat. For this reason the flat response is now the default target for all standard configuration files. The standard target response stage should be used only to adjust the response to taste, but, unlike previous versions, for a neutral reproduction a flat response should be used.
From the subjective point of view a system equalized to the inverse of the magnitude response envelope usually sound really neutral. Even though most of the times the magnitude response envelope response is different among the various channels, resulting in an obvious channel misalignement if evaluated with the standard smoothing procedures, imaging usually improves, becoming more stable and focused. This appear to confirm that the estimation performed by the magnitude response envelope has to be close to the subjective perception of the magnitude response.