It’s personal preference. Slow filters are often better for small jazz groups, solo voice, etc. For bombastic music or dense orchestral music I prefer brickwall. Some discs will definitely benefit from an apodizing filter to get rid of filtering artifacts from previous upsampling, etc. Linear phase filters preserve the waveshape better than minimum phase filters, but some like the extra energy from (some) minimum phase filters.
I’m using slow filters both linear phase and minimum phase and hearing such differences. I thought the recessed vocal and more spacious sound has something to do with the minimum phase and the more forward vocal and narrower soundstage has something to do with the linear phase filter.
Again, i am so new to this I need all the help I can get.
I’m not sure you’ll get uniform experiences with the different filters. There’s no mathematically perfect filter and for every filter type there are good and bad implementations. I like linear phase filters because they most faithfully reproduce the incoming waveshape. But for many kinds of music that waveshape has already been distorted a lot and perhaps faithful reproduction isn’t the most important thing. Reproducing the waveshape is perhaps most important for transients, percussive sounds.
I don’t doubt what you hear, but for me the differences aren’t related so soundstage in general. It’s more likely to do with the particular music you are playing on each filter or perhaps on the implementations of the filters in your DAC.
Anyway, you’ll probably gravitate to a particular choice over time and forget that you can even play with the filter type
Historically many of the filters in earlier DACs and upsampler chips used things like coefficients with (figuratively speaking) few ones or zeros or as few coefficients as p0ossible. These days they can afford better filters, but they don’t always do that. If you want really good filters, they take a lot of coefficients which is a lot of memory so having many choices can be expensive. Putting those coefficients outside of the FPGA means you need extremely high bandwidth to read all of the ones needed for each incoming sample… Arguably having a choice for an apodizing filter seems like a possible compromise. Perhaps having loadable filters would make some happy, but then it might take more time to change selections.
Interesting information… I’m doing these comparisons with the same cd and tracks to baseline what I’m hearing. My cd player is a Mark Levinson 5101 which is their newest model with their new DAC (in hindsight I would have kept it all PSaudio). If I may ask, since I’m leaning towards the Linear Phase filter because of vocal clarity, in layman’s term, what am I looking for between the Slow Linear vs the Fast Linear?
Slow and fast refer to how fast the transition is between the pass band and the stop band. With a fast transition you get more highs (tho perhaps only a few notes) and with a slow transition you may lose those highs but gain less distortion in the phase of the high frequencies. With good upsampling this is much less of a problem than it was in the past. Still, many prefer slow transitions for simpler music, not a lot of instruments or voices. On the last DAC I had where I had a choice, I just tried them without paying attention to their labels, each for a week or two until I found the one I liked best overall. No surprise to me but it was the brickwall filter which is basically as fast of a transition as possible. But not all DACs implement brickwall filters well so you may indeed like another filter better on your DAC.
Brightest isn’t the way I’d describe it. With large orchestra or, say, Alice In Chains it just seems more natural to me. It doesn’t artificially roll off the top like fast or slow transition filters. If the brickwall filter is implemented well, it also doesn’t have any aliasing that some filters have. So a brickwall filter may indeed be less bright than others if those others allow a little leakage from above the Nyquist rate (which some slow filters do allow.) Some people like the added energy of aliasing, for example it makes transients more apparent - that’s not my taste, but there’s nothing wrong with liking it.
Since most DAC chips have provided DAC filters for a long time many CD players expose them thru the user interface. My SCD-777ES from twenty years ago had them and there were plenty of others before that.
When they were implemented badly it didn’t help the sound of the DAC. As time has gone on the upsampling filters in DACs have gotten better and better. Many early filters aliased and that gave early CD players an edgy sound. But it was also an “exciting” sound that some preferred to vinyl. I still have some friends that prefer the sound of older DACs.
Analytic is/was used pretty freely in the past to describe a dry non-involving sound with lots of details at the expense of involvement or perhaps comfort. The thing is that there are many ways to screw up sound and end up there.
There’s a sound that a lot of people called analytic which was really just a slight tilt up at high frequencies. That let’s you hear more detail, but (for most of us) gets tiring fairly quickly. I was quite surprised at how much detail could be found with a flat frequency response. I really didn’t want to mess with finding the best frequency response shape for multiple customers.
Letting some aliasing thru adds high frequency content that isn’t harmonically related to the source signal. Some hear that as extra energy, some call it a headache. Some called it analytic.
Your phrase
seems to be leading to me: I don’t agree that DACs are more or less analytic or that being analytic is intrinsic in being a DAC. I do agree that many DACs can be analytic, but also many systems without DACs can be analytic. At times the best system I’ve ever listened to was analytic due to an (IMO) unfortunate choice of interconnects.
