INTERACT FORUM

More => Old Versions => JRiver Media Center 23 for Windows => Topic started by: paoloboccardi on January 13, 2018, 08:30:15 am

Title: Convolution and resampling
Post by: paoloboccardi on January 13, 2018, 08:30:15 am
Hi, when I use a 96hz filter in convolution plugin, I need resample all my audio file to 96hz (with dsp plugin)?
Thank you
Title: Re: Convolution and resampling
Post by: Trumpetguy on January 15, 2018, 08:56:02 am
Yes. Use the MC resampler, works really well.

My experience is, however (after resampling to 96kHz for years), that creating convolution filters for the most common sample rates gives better results. JRiver picks the filter (there are naming rules for the filter config files that makes this possible, look in the wiki) that fits with the source material.
Title: Re: Convolution and resampling
Post by: paoloboccardi on January 15, 2018, 10:45:52 am
Yes. Use the MC resampler, works really well.

My experience is, however (after resampling to 96kHz for years), that creating convolution filters for the most common sample rates gives better results. JRiver picks the filter (there are naming rules for the filter config files that makes this possible, look in the wiki) that fits with the source material.

Yes I think is better solution: In my case, the UMIK microphone measures well at 44, but at 96 it inserts a lot of noise. I could create a 44 filter and do upsampling to 96 with an external program. In this case I have to check the normalization function in jriver, correct?
Title: Re: Convolution and resampling
Post by: Trumpetguy on January 15, 2018, 03:54:54 pm
You don't actually have to do separate room response measurements for all sample rates. You can do one sweep at one rate and have resample the room response measurement and make one filter for each (using e.g. Audiolense makes this step real easy). For some reason, this sounds better in my system than hving one filter and resample all audio in real-time.
Title: Re: Convolution and resampling
Post by: Trumpetguy on January 15, 2018, 03:57:41 pm
Also, I find doing separate room response measurements for multiple sample rates really difficult. They always have individual characteristics, probably because it practically impossible to get 100.0000% equal conditions fro two consecutive measurements. I find the filters to be more consistent (and way less time consuming to create) by having Audiolense save filters at all necessary sample rates based on one measurement.

Edit: some wording
Title: Re: Convolution and resampling
Post by: paoloboccardi on January 16, 2018, 05:59:41 pm
Thank you for answer, I will try with R8brain pro: this software is capable to resample different type of filter. Usually with 1 filter and resample on fly, I do not use the normalization option for gain, but I use an equalization by inserting -3 gain for a 44 filter to avoid 100% peak. Do you automatically normalize?