Sound Forge i noise reduction

[Marko37]

Odvoji sam wav od nekog mpeg-2 skinutog sa videa koji je pun shuma. Video signal sam prilicno "ocistio", ali do sada nisam radio sa SF i NR pluginom, pa ako neko moze da mi da par saveta oko najboljeg nacina za filtriranje shuma.
LM, signal je takodje problematican. Desni je ok, ali levi pocinje skoro bez ikakvog signala, pa do kraja izuce oko 30-40% od signala desnog. Da li se isplati da filtriram i normalizujem levi, tj. da zadrzim stereo, ili da prekopiram desni i napravim mono signal? Koji bi redosled bio efikasniji: filtriranje pa normalizacija ili obrnuto? Po mojoj logici trebalo bi da je svejedno, ali...
BTW da li mogu da napravim spajkovito snazno filtriranje na 50Hz i celobrojne umnoske, a da ne diram ostatak spektra?

 

[jddipqd]

Koji bi redosled bio efikasniji: filtriranje pa normalizacija ili obrnuto? Po mojoj logici trebalo bi da je svejedno, ali...

Prvo filtriranje pa onda normalizacija. Ako prvo normalizuješ pojačaćeš i šum.

 

[diviner]

Svejedno je ako koristis staticki noise reduction. Tada se radi relativno, u odnosu na signal pa ako ga i sve zajedno pojacas on ce ga "zagusiti" na tim frekvencijama (koje izaberes), u onom odnosu u kom mu navedes (procetnualno)!
Potreban ti je dinamicki noise reduction plugin (neki bolji...pogledaj postove , neko se vec doticao toga) ili ...recimo Steinberg Clean:

Mozda ce ti biti ovaj text od pomoci:

You can reduce noise in existing recordings in a variety of ways, some of which are more effective than others. Expansion (the opposite of compression) can widen the dynamic range, and normally in the case of noise reduction the majority of the signal passes through unaltered -- only when the signal level drops below a user-set threshold does expansion occur. This technique can be effective, although as with compression, it can give rise to pumping effects if over-used. Another way to reduce noise is to use a variable low-pass filter such that the cutoff frequency is reduced during quiet passages, although this time over-use can result in deliberately low-level passages becoming muffled. A better way is to analyse the audio by separating it into various frequency bands, and then expanding them individually. Generally, the more bands in the analyser algorithm, the more expensive the product, and the more effective the results, but this is not always the case.
Most noise-reduction software finds it easier to distinguish background noise from music if you can first feed it a short section containing just noise (which can often be found between tracks on cassette or vinyl). The software can then use this as a noise 'signature' (or noise print), and remove noise with the same frequency spectrum as the signature from the entire recording. However, Clean! uses a similar noise-reduction algorithm to the much more expensive plug-ins designed by Spectral Design, which works in real time, constantly adapting its noise print during the course of the track. More than five years of research went into the original algorithms, and it shows, even on the cut-down versions included in Clean!

POZDRAV!

 

[Marko37]

Re: Re: Sound Forge i noise reduction

Prvo filtriranje pa onda normalizacija. Ako prvo normalizuješ pojačaćeš i šum.

U realnom svetu redosled ne bi trebao da ima znacaja jer uspeh sklanjanja valjda zavisi od SNR (signal2noise ratio), tj. od odnosa signal/shum.

 

[zagortenej]

ljudi bolje prvo normalizujte - savet iz glave - nisam probao ali po mojoj teoriji to je bolje jer su to digitalni signali - da su analogni ne bi imalo veze