- Joined
- Sep 4, 2019
- Messages
- 1,172
- Likes
- 2,322
I have hit the limit of my knowledge and in researching have been unable to come up with answers. Fortunately, there are those here who are far more knowledgeable, so I am hoping someone can provide the answers. Here it goes:
I have calibration levels set for both my systems for reference level listening using Bob Katz's recommendations for mastering and adjusting for room size as suggested. I employ volume leveling in Roon and Jriver at -20 dbfs in order to have different level recordings play back at the same level. In the vast majority of cases, I see that Roon or Jriver reduce playback level by 5-12db with volume leveling (I listen to a lot of studio recorded music, so there is varying levels of compression).
In the case of my desktop, I calibrate to 74 dbc (each speaker so 77db for the pair) using 500-2,000hz -20dbfs band limited pink noise. For my living room it is 80dbc for each speaker (I have a pretty large room). My understanding is that by using this calibration level, the highest level I should ever see on my main speakers in the 500-2000hz range is 20 db above the calibration level for a 0dbfs signal, since a digital signal cannot go louder. This would translate into 97db for the pair of speakers in my desktop system, which just happens to have a listening distance of 3 feet (or one meter for the majority of the world).
Here is what I don't understand. When using my SPL meter, whether in REW with a UMIK or on my phone, I see average levels right where I expect. I also see Lmax levels, right where I would expect. What I don't understand is how my LC peak levels can be so much higher, what I would expect to be way above what is possible without digital clipping. Lets take a specific example:
This is a snapshot of a 1 minute segment of Chilly Gonzales- Sample This from Live at Massey Hall, which has a loud short peak at about 2:40 in the track. In Roon, volume leveling has reduced it by 10.6db to hit my target -20dbfs. I am using fast A weighting to eliminate any SPL contribution by my subwoofer which is set a bit above the mains level. As I would expect, LAeq is right at 77db. I would expect the max SPL possible to be 77db + 9.4db (because volume has been reduced by 10.6 db for volume leveling) or 86.4 and my Lmax is about as expected at 84.4 db (understanding that even with Lmax for fast weighting it is still an average level, but over an extremely short time). Here is what I don't understand:
Thanks for what I hope is an informative discussion.
I have calibration levels set for both my systems for reference level listening using Bob Katz's recommendations for mastering and adjusting for room size as suggested. I employ volume leveling in Roon and Jriver at -20 dbfs in order to have different level recordings play back at the same level. In the vast majority of cases, I see that Roon or Jriver reduce playback level by 5-12db with volume leveling (I listen to a lot of studio recorded music, so there is varying levels of compression).
In the case of my desktop, I calibrate to 74 dbc (each speaker so 77db for the pair) using 500-2,000hz -20dbfs band limited pink noise. For my living room it is 80dbc for each speaker (I have a pretty large room). My understanding is that by using this calibration level, the highest level I should ever see on my main speakers in the 500-2000hz range is 20 db above the calibration level for a 0dbfs signal, since a digital signal cannot go louder. This would translate into 97db for the pair of speakers in my desktop system, which just happens to have a listening distance of 3 feet (or one meter for the majority of the world).
Here is what I don't understand. When using my SPL meter, whether in REW with a UMIK or on my phone, I see average levels right where I expect. I also see Lmax levels, right where I would expect. What I don't understand is how my LC peak levels can be so much higher, what I would expect to be way above what is possible without digital clipping. Lets take a specific example:
This is a snapshot of a 1 minute segment of Chilly Gonzales- Sample This from Live at Massey Hall, which has a loud short peak at about 2:40 in the track. In Roon, volume leveling has reduced it by 10.6db to hit my target -20dbfs. I am using fast A weighting to eliminate any SPL contribution by my subwoofer which is set a bit above the mains level. As I would expect, LAeq is right at 77db. I would expect the max SPL possible to be 77db + 9.4db (because volume has been reduced by 10.6 db for volume leveling) or 86.4 and my Lmax is about as expected at 84.4 db (understanding that even with Lmax for fast weighting it is still an average level, but over an extremely short time). Here is what I don't understand:
- How can I be seeing LCpeak of 102.3 db? My calculations show this is nearly 16db above 0dbfs.
- When we discuss the peak power needed to avoid clipping an amplifier, shouldn't we be discussing LCpeak levels? This is a huge difference where in the one case, my speakers need about .5 watt each to hit 85 db (keeping in mind I am measuring a pair, individually each would be 3 db lower), the other 32 watts. Increase my reference listening level to 83 db and these peaks are asking for 128 watts per channel, getting very close to the 150w/ch/8ohms my amp puts out.
Thanks for what I hope is an informative discussion.