Average music power of most popular music is a LOT lower than you might think

[mike7877]

For example, the metal genre (which usually exists in good proportions over the entire audio spectrum and for this reason is also a good representation of all music {except the very bass heavy genres which can vary widely}) generally is just 7 watts per 100

That's right! Just 7 watts average, with peaks at 100!

And those peaks are the consistent peaks - not the main peak. Most new music is compressed (amplitude-wise) so the largest peak is now usually brought down to the level of the main peaks, instead of it running 3-4 (or more) dB above them.


With very revealing speakers and good recordings of most popular music (stuff that stays at about the same amplitude from start to finish,) you can get to maybe 10 without affecting sound and dynamics too too much. 12 watts for lesser systems. Listening fatigue will set in sooner with distortion though (and usually if you're at your amp's limit and it's not undersized, listening fatigue would be setting in pretty quick anyway). Maybe you'd get 10 minutes instead of 15.


If you didn't know this already, reply with what you thought the ratio was. This ratio (average : peak) is:

1:15

 

[ahofer]

I think I'm missing some context.

 

[fpitas]

Classical music peak-to-average can get up to and beyond 30dB. Regular pop doesn't usually break 15dB.

 

[Gary_G]

I started to think abou this before my final purchases. Information here and other sites lead me to looking for more efficient speakers to pair with an amplifier. I like the pairing of the Parasound A23+(160watts) amplifier and P6 preamplifier and finally combined them with Focal Aria 926 speakers. I have a system that doenst have to work hard at all.

 

[fpitas]

Power calculator on this page, if you don't want to run through the math:

Crown Audio - Professional Power Amplifiers | English (US)

www.crownaudio.com

 

[Blumlein 88]

Yeah, the decibel difference between 7 and 100 watts is not quite 12 decibels. So not surprising, and you'll find reducing power to 10 watts not fully satisfying.

 

[kongwee]

With crossover and natural induction of conventional driver, you amp will draw more current. Your speaker can output 1watt at 90dB, you amp may have to push 2ampere. Let say your transistor is rated 12V max. 2 x 12 =24 VA. Under pure resistive load is 24watt max. But you speaker only need 1watt, the rest of the 23VA will become heat lose. Some amp spec will slam current output 10 or 16ampere which is very high.

 

[Beave]

With crossover and natural induction of conventional driver, you amp will draw more current. Your speaker can output 1watt at 90dB, you amp may have to push 2ampere. Let say your transistor is rated 12V max. 2 x 12 =24 VA. Under pure resistive load is 24watt max. But you speaker only need 1watt, the rest of the 23VA will become heat lose. Some amp spec will slam current output 10 or 16ampere which is very high.

Can somebody explain this one? Amps drawing current and speakers outputting watts? Transistors rated at 12V max? Is this an alternate universe?

 

[ebslo]

I was curious, so I checked my library. Over 3004 flacs with ReplayGain, the average REPLAYGAIN_TRACK_GAIN is -4.5dB. IIRC, replaygain, in a roundabout way, normalizes to -14dB reFS, so the average "loudness" by it's measurement method is -9.5dBFS with a standard deviation of 3.6dB. This was after removing one outlier aptly named "silence".

For those that like power ratios, -9.5dB is about 1:9.

 

[kongwee]

Can somebody explain this one? Amps drawing current and speakers outputting watts? Transistors rated at 12V max? Is this an alternate universe?

It is about Alternating Current or Voltage. It is a different universe from Direct Current or Voltage. AC and DC to simplify wordings.

 

[Beave]

It is about Alternating Current or Voltage. It is a different universe from Direct Current or Voltage. AC and DC to simplify wordings.

I agree with the "alternating...universe" part of your statement.

 

[DVDdoug]

I was curious, so I checked my library. Over 3004 flacs with ReplayGain...

Replay gain doesn't tell you anything about the peaks or crest factor. Except the developers made some assumptions & compromises related to dynamics so that most tracks can be adjusted to the target without clipping.

 

[DonH56]

IME most people overestimate their average power and underestimate the peaks. Many years ago a study I have long since forgotten presented a peak to average ratio of 17 dB for music, a power ratio of 50:1. That was well before CDs. More recently I have read up to 20 dB (100:1 in power) for some music recordings, and up to 30 dB (1000:1 in power) for movies (explosions, gun shots, and so forth).

The power you use depends upon your speakers' sensitivity, position in the room, and distance to the listener (among other things). Here is an online calculator that can help estimate power need and usage: http://myhometheater.homestead.com/splcalculator.html Note bass may require much more power; see equal loudness curves.

1 dB change is hardly noticeable and requires 1.26x the power (about 1/4 more).
3 dB change is "a little bit louder" and requires twice (2x) the power.
6 dB change is noticeably louder and requires 4x the power.
10 dB change sounds twice as loud and requires 10x the power.
17 dB is the peak-to-average range of most music and is a 50:1 spread in power.
20 dB is four times as loud and requires 100x the power.

HTH - Don

 

[Cars-N-Cans]

Can somebody explain this one? Amps drawing current and speakers outputting watts? Transistors rated at 12V max? Is this an alternate universe?

I think he means for linear amplifiers. There, the transistors just act as basically variable resistors that control how much current is drawn from the power supply. If your speaker has a varying impedance with a minimum of about 2 ohms, at 2.83V this is 1.4A as opposed to 0.35A with an 8 ohm resistive load. If you have a linear amplifier with +/- 36V power rails, the voltage drop across the output transistors will be 36V - 2.83V, or about 33V. This all becomes heat. In the case of the speaker with a 2 ohm minimum impedance, the instantaneous power dissipation inside the amplifier will be 33V * 1.4A, or 46W. With a purely resistive 8 ohm load, the dissipation will be a more modest 12W instead. This is why class-D amps are better in general, as the PWM switching and output networks allow a better impedance match to speakers with much reduced heat dissipation compared to linears where there is no impedance match, and about half of its output is wasted as heat.

 

[ebslo]

Replay gain doesn't tell you anything about the peaks or crest factor. Except the developers made some assumptions & compromises related to dynamics so that most tracks can be adjusted to the target without clipping.

Statistics for REPLAYGAIN_TRACK_PEAK over the same collection: mean = 0.901, standard deviation = 0.145. These are not dB, just peak / full scale. So that adjusts my previously reported -9.5dB reFS mean "loudness" (as determined by the model replaygain uses) to -8.6dB below mean peak.

 

[Beave]

I think he means for linear amplifiers. There, the transistors just act as basically variable resistors that control how much current is drawn from the power supply. If your speaker has a varying impedance with a minimum of about 2 ohms, at 2.83V this is 1.4A as opposed to 0.35A with an 8 ohm resistive load. If you have a linear amplifier with +/- 36V power rails, the voltage drop across the output transistors will be 36V - 2.83V, or about 33V. This all becomes heat. In the case of the speaker with a 2 ohm minimum impedance, the instantaneous power dissipation inside the amplifier will be 33V * 1.4A, or 46W. With a purely resistive 8 ohm load, the dissipation will be a more modest 12W instead. This is why class-D amps are better in general, as the PWM switching and output networks allow a better impedance match to speakers with much reduced heat dissipation compared to linears where there is no impedance match, and about half of its output is wasted as heat.

That's a very, uh, magnanimous interpretation of his confusing post. Yours makes sense. His did not.

 

[kongwee]

Impedance is never a pure resistive load. Even with DSP amplifier will draw more current in each driver. Each driver(except for electrostat) are inductor itself with it's internal resistance. This lower frequency the inductor will draw more current than higher frequency. Your woofer may output 1 watt at 90db at 60Hz, it may need 2av(voltage x current in ac). In case it is a power factor of 0.5 or phase angle at 90 degree lagging. It may need 0.5 watt at 90 at 120Hz, it may need less than 1va. In any case, you can't avoid extra current drawn due to inductor. That why in your speaker impedance graph, it is never linear. Sometime some amplifier will just give current output rating beside the power rating. I should extra current drawn for nothing and create internal heat lossses. A dump load in amplifer to measure never represent the real world crossovers with drivers. Impedance topic is not easy to understand w/o engineering background.

 

[DonH56]

The crossover has a major impact on the impedance the amplifier sees.

 

[sarumbear]

For example, the metal genre (which usually exists in good proportions over the entire audio spectrum and for this reason is also a good representation of all music {except the very bass heavy genres which can vary widely}) generally is just 7 watts per 100

That's right! Just 7 watts average, with peaks at 100!

And those peaks are the consistent peaks - not the main peak. Most new music is compressed (amplitude-wise) so the largest peak is now usually brought down to the level of the main peaks, instead of it running 3-4 (or more) dB above them.


With very revealing speakers and good recordings of most popular music (stuff that stays at about the same amplitude from start to finish,) you can get to maybe 10 without affecting sound and dynamics too too much. 12 watts for lesser systems. Listening fatigue will set in sooner with distortion though (and usually if you're at your amp's limit and it's not undersized, listening fatigue would be setting in pretty quick anyway). Maybe you'd get 10 minutes instead of 15.


If you didn't know this already, reply with what you thought the ratio was. This ratio (average : peak) is:

1:15

Without correlating SPL values what you are saying does not mean anything, I’m afraid.

 

[fpitas]

Without correlating SPL values what you are saying does not mean anything, I’m afraid.

It is admittedly very open-ended. But a good subject to educate people about why they need such a powerful amp if they listen at say, 1W average.