The harshness...
The RIAA Preamp seemed the most 'unpleasant' sounding in the sample from around 14'45" into the track. The voice tremolo became more of a wah-wah that seemed to get 'inside my head'. The IIR min phase seemed to keep things together better although dare I say that it sounded a little more constricted on the vocal than the RIAA but overall more pleasant. The FIR linear phase seemed to have 'a little of both' but overall I would take the IIR minimum phase version (again
)
The RIAA Preamp seemed the most 'unpleasant' sounding in the sample from around 14'45" into the track. The voice tremolo became more of a wah-wah that seemed to get 'inside my head'. The IIR min phase seemed to keep things together better although dare I say that it sounded a little more constricted on the vocal than the RIAA but overall more pleasant. The FIR linear phase seemed to have 'a little of both' but overall I would take the IIR minimum phase version (again
)
I remember the first time in a new workshop (big Norweb depot) and suddenly there was this loud rumble. None of us knew what it could be but it was really alarming. Same again the next day, I went outside and all was quiet. Back indoors and there it was. Quizzing some older staff around the place and it turned out there was a railway line (used by Royal Mail apparently although it connected to the main networks) that was directly under our building. Looking at street maps and you could see the lines and where they disappeared underground, some for half a mile or more.
And today, even using my better headphones I can't hear it. I'll break out the electrostatics tonight. Three hours of conference calls with a headset on probably doesn't help hearing accuracy
I'm sure
I've been listening on Fostex headphones and do not hear it. Maybe my left ear is shot? I'll try swapping the phones.
Also, between the analog and the digital versions, the singer's voice is more centralized on one than the other. One of the versions spreads it across let and right more than the other. It's easy to notice on headphones. Does anyone else notice this between the analog and digital filters?
EDIT: I think the singer is Hoffman. Sounds Mezzo to me.
Also, between the analog and the digital versions, the singer's voice is more centralized on one than the other. One of the versions spreads it across let and right more than the other. It's easy to notice on headphones. Does anyone else notice this between the analog and digital filters?
EDIT: I think the singer is Hoffman. Sounds Mezzo to me.
The masochist Mahler has turned post mortem into a sadist
The digitals have more well defined and stable centre image than the analog.
It’s not only with the mezzo, it’s with the brass wind instruments too. And if you ask me, the FIR minimum phase is also the sharpest sounding
George
The overall result seems to be coming down in favour of the digital processed ones.
Sounds like you might be noticing a little of the same kind of thing that I did. For me it was the RIAA version that did this and I didn't like it as much as the others.
The overall result seems to be coming down in favour of the digital processed ones.
Thanks for doing this guys, unfortunately I have not had the time to listen to these. Kinda got a little burnt out doing the article.
When doing the article I only listened to IIR via miniDSP using a Tascam DR70 as preamp and FIR in Foobar with foo-convolve from the Tascam 24/96 files. I used a couple of RCA/JVC Italian film scores that were particularly well recorded and only had 5 or 10 plays on them. I'm stuffing a new analog RIAA so I might repeat the exercise.
Last edited:
And thank you Scott for doing the leg work behind this. I am looking forward to the point where I eventually get the miniDSP into my system and start playing with this.
@George: have you considered running one of the digital files through the declicker Scott posted a while back to see what effect it has ?
@George: have you considered running one of the digital files through the declicker Scott posted a while back to see what effect it has ?
BTW I found it useful to do one pass by hand to do the worst huge clicks one at a time. If you saw my picture of almost 5 or 6 cycles of a 1kHz test tone being almost perfectly replaced, I later discovered that doing a larger length selection did not do as well on the occasional huge pop (at least it looked that way). The Clickfix plug-in and CoolEdit 2000 still work fine on my new W10 laptop (sans ASIO though).
OK, risking wrath of Memsahib (long story) I put the ESP/950s on for another listen. Much better . Now listening again, at 8:59 you can clearly pick up the tube rumble but I cannot for the life of me pick up that beeping again. Slight cold so my left ear feels slightly blocked so put the headphones on back to front... Still nothing. This wouldn't bother me had I not heard it with a pair of cheapy earbuds last night. will keep at it.
. Now listening again, at 8:59 you can clearly pick up the tube rumble but I cannot for the life of me pick up that beeping again. Slight cold so my left ear feels slightly blocked so put the headphones on back to front... Still nothing. This wouldn't bother me had I not heard it with a pair of cheapy earbuds last night. will keep at it.
Yes, that is what I hear. I don't hear any difference in image between the digital versions, but between analog and digital, it's noticeable. Like maybe an interchannel phase shift in the analog RIAA.
Seems that Mooly is hearing the same thing.
Hmmm, I was going to say that in my first impressions, but I could only reliably notice it on the ticks and pops. FIR is sharpest. Not sure I could pick it out on the music.
We need to be careful in characterizing the complete signal path for each before making any definitive conclusions.
I don't recall so the safest bet is to assume no. We also need to verify the math precision each tool uses, I mentioned some problems in my article like at 32bit the order of IIR filters can matter by quite a bit. One stake in the ground could be to apply each digital technique with verifiable 64 bit double precision math on exactly the same non-RIAA rip from the vinyl. Python and SoX always use 64 bit unless directed otherwise.
these kind of things all add to the ambient background atmosphere.OK, risking wrath of Memsahib (long story) I put the ESP/950s on for another listen. Much better
Our hearing is very variable from day to day (so I believe). You'll pick it up again
- Status
- Not open for further replies.