The issue of digital vs analogue amplifier is more or less settled for me, and I could actually hear the difference on one of our own members YouTube video, with the digital amp sounding harsh and 'clinical' I call it that I suppose because it reminds me of surgical instruments being dropped on an operating theater floor. In fact I purchased a 'Lepai' type digital amp and was very disappointed with it, to the extent I prefer an amp salvaged from PC speakers and the LM386 board. To be fair, I have compared the output with that of a vintage Sony amp and speakers and they compare favorably, about maybe 70% of the quality.
It is difficult for me to understand why DACs sound different, and also why DVD player DACs sound not as good as a modern dedicated DAC. Is it speed, accuracy, error correction? Might need to see the threads on how DACs work.
Did that, sounds clearer - so -3 dB is to lower the max amplitude to leave enough headroom for EQ to act, and then normalize to 0 dB to get the max amplitude again.
This applies to .mp3 as well as .wav files or only .wav?
I don't want to follow the audiophool route of flowery language since I'm an engineer, but please excuse any inevitable descriptions and words!
I only tried the digital vs analogue power amp on the HF drivers in a PA system, and in addition to the slight harshness which you also noted, much of the realism vanished - the sound simply sounded more artificial with the instruments less distinctly separated. (Here I go!). A visual analogy would be like comparing a sharp hi-res image printed on glossy photo paper with a slightly out-of-focus version printed on ordinary office print paper. One could see exactly what the image contained in both instances, just that one was nicer that the other.
I really don't know where to start with DACs given the spectacular on-paper performance of even the cheapest offerings when compared with that of even very good loudspeakers. (THD, Intermodulation distortion etc.). It's one of those intangibles which allows buyers to justify spending huge sums of money, aiming for their idea of perfection. All I will say is that I bought a used £100 Cambridge Audio DAC which was considerably easier on my ears than a similar age £60 Chinese one it replaced and which I bought new a few years back.
Good luck with your quest.
It applies to all* - and if you're putting more than 3db on it, normalized to -6db or whatever.
*When going to mp3, I usually normalize to -0.5db to give room for the encoder to play with - if you encode a 0db file with mp3, it will clip.
The issue with that is the EQ response given above shows 10dB gain or so at some frequencies, so there can still be clipping, but this will depend on the actual signal.
I think a more logical approach is some EQ after the CD player and before the amplifier - much less work in the long run, as the EQ is to match the loudness level and the room acoustics, not really anything to do with the recording medium. This avoids the levelling issue completely.
I am using a DVD player and its internal DAC: since you have heard some differences in the sound of DACs, how does a DVD player sound compared to a dedicated DAC, both the DACs that you bought?
My DVD player is this one: (or similar)
https://www.usa.philips.com/c-p/DVP3620_F7/dvd-player/
Regrettably I am unable to comment on DVD players as my sources are either the PC or the digital out from an antique £35 technics CD player. I have only compared my DACs with each other and a mid-priced CD player. The spec. on paper of your machine is fantastic, as are cheap Chinese DACs; where the 'magic' comes from in more expensive DACs I have no clue! So far the Cambridge DAC is, to me, the best sounding of the lot, but in a subtle way. It's certainly not a night-and-day difference, simply easier on the ear and a little more detailed and open-sounding, but the new (used!) £160 headphones remain the biggest improvement by far as I expected they would be. It is likely that the biggest improvement in your listening pleasure would be achieved with new speakers, but if like me there is a limited budget then this would be impossible. Your EQ experiments are costing you nothing and will improve the subjective sound considerably, and a different DAC need cost no more than £150 or so. Another free option is to experiment with speaker placement, and it would not cost much to add some room treatment to improve the acoustics - these two options probably remain your cheapest overall route to improved sound. I would also suggest room EQ using something like the cheap and versatile Behringer DEQ2496 or even cheaper but basic FBQ2496.
Good luck, Carl.
Thanks. Two facts need mentioning: My speaker are not going to move further into the room because of concerns regarding the appearance of the room.. you know what I mean.
The other fact is that I have not heard a high end audio system such as the ones many of the enthusiasts here have at home, never in my life. There are no audio shows where I live, and no one I know has high - end stuff. I can only guess from what I hear on You Tube.
Maybe someday... but that's the attraction of the hobby - it's the journey.
The other fact is that I have not heard a high end audio system such as the ones many of the enthusiasts here have at home, never in my life. There are no audio shows where I live, and no one I know has high - end stuff. I can only guess from what I hear on You Tube.
Maybe someday... but that's the attraction of the hobby - it's the journey.
I normalize after EQ. Since the editor works in 32 bit float, there isn't any risk of clipping while editing. But it will be clipped at export if the level hasn't been brought down. IIRC Audacity normalize is default -1dB which is fine for PCM. Maybe a little more is best for MP3? Haven't tested that.
I use Audacity in 24 bit mode so I reduce the level before I eq...
If I rip a CD for encoding, I normalize to -0.5db so the encoder doesn't clip. I don't use MP3 anymore but old habits die hard 🙂
If I rip a CD for encoding, I normalize to -0.5db so the encoder doesn't clip. I don't use MP3 anymore but old habits die hard 🙂
OK, AFAIK all processing is done in 32 bit float. Is there a 24 bit integer working mode? I use Goldwave more often, and it's float - so no worries. I've done the the test in Audacity, boosting way above 0dB so that it looks clipped on the waveform, but in fact it isn't clipped at all. You can reduce the volume and there is no clipping. However if you save into 16 or 24 bit PCM it will certainly be clipped! That's the nice thing about 32 bit float - it has something like 500 dB dynamic range, making it easy to stay away from clipping.
For a long time I was super cautious in wave editors never to clip - then I found out that the signal wasn't clipping at all, just the display of the waveform was clipping.
For a long time I was super cautious in wave editors never to clip - then I found out that the signal wasn't clipping at all, just the display of the waveform was clipping.
Audacity can do batch processing.
https://docs.huihoo.com/audacity/2.0.3/man/batch_processing.html
Looks like you have to set up the "chain" of effects, then apply it to a batch of files.
For example: Chain = Apply "LIVINGEYES22" curve preset. Normalize to -1dB. Export to wave 16 bit, 44.1Khz stereo
https://docs.huihoo.com/audacity/2.0.3/man/batch_processing.html
Looks like you have to set up the "chain" of effects, then apply it to a batch of files.
For example: Chain = Apply "LIVINGEYES22" curve preset. Normalize to -1dB. Export to wave 16 bit, 44.1Khz stereo
I remember Goldwave! I haven't used it since the 90's! I switched to Cool Edit and then Audition before I ditched Windows.
I'm on Manjaro Linux these days but I hear GoldWave works through WINE...
I was building Audacity from git but I went to an appimage of the latest stable because the alpha is broken and the audacity team can't be bothered to fix their packaging errors so Arch and it's derivatives are on 2.4.1 instead of 3.1.3.
I'm on Manjaro Linux these days but I hear GoldWave works through WINE...
I was building Audacity from git but I went to an appimage of the latest stable because the alpha is broken and the audacity team can't be bothered to fix their packaging errors so Arch and it's derivatives are on 2.4.1 instead of 3.1.3.
I'm super comfortable in Goldwave because I've been using it for about 25 years. I run it on Windows, and have done it on Wine on the Mac. Goldwave is working on a new version that runs in a browser, so will work on most any OS.
For the Audacity batch thing I went to have a look and on my copy on Mac OS there is no effects chain. Damn. So I have downloaded and will install the latest version to see if it includes the effects chain and batch files commands.
For the Audacity batch thing I went to have a look and on my copy on Mac OS there is no effects chain. Damn. So I have downloaded and will install the latest version to see if it includes the effects chain and batch files commands.
An mp3 can't be 'fixed' with EQ. The problem with it is that it is missing information which was thrown away to make the file smaller. The lost of information is permanent.
That said, there are DSP effects that can be used to 'band-aid' some types of problems found on poor recordings. For example, there are harmonics synthesizers that can sorta simulate missing HF information.
Lost punch is another matter. Sometimes a slow-attack compressor can be adjusted to allow the initial part of dynamics transients through, then compress a short time later. So long as the release time is set to that compression effect has decayed before the next transient occurs, a sense of punch can sometimes be produced. Works best if the transients are evenly spaced in time, such as can be found in the hits of an exceptionally good drummer playing dance music.
Best thing is to avoid lossy compression in the first place if you don't like the sound of it. If lossy compression is unavoidable, AAC is a newer lossy compression algorithm that is better sounding than mp3.
That said, there are DSP effects that can be used to 'band-aid' some types of problems found on poor recordings. For example, there are harmonics synthesizers that can sorta simulate missing HF information.
Lost punch is another matter. Sometimes a slow-attack compressor can be adjusted to allow the initial part of dynamics transients through, then compress a short time later. So long as the release time is set to that compression effect has decayed before the next transient occurs, a sense of punch can sometimes be produced. Works best if the transients are evenly spaced in time, such as can be found in the hits of an exceptionally good drummer playing dance music.
Best thing is to avoid lossy compression in the first place if you don't like the sound of it. If lossy compression is unavoidable, AAC is a newer lossy compression algorithm that is better sounding than mp3.
Last edited:
Regarding recordings with different perceptual volume level, even after normalization, the louder ones have probably been run through some combination of compression (possibly multi-band) and or limiting. One option might be to process the low perceptual level recordings to make them sound louder (increasing the average volume level rather than the peak volume level). Another option could be to lower the volume level of the louder recordings to match the softer ones. To help with that there are now 'perceptual VU meters' that can be used to measure perceived loudness. Some may even be able to renormalize to some target perceptual volume level. Maybe some tool like: https://www.tbproaudio.de/products/deq6 ...Or maybe: https://youlean.co/youlean-loudness-meter/
I'd just like to interject and say that - any day of the week - I would rather listen to well-recorded music on lossy MP3 than a 'perfect' FLAC file of a bad master. I feel that many people really get too hung up on the source encoding method to see the wood for the trees. Most listeners, myself included, would be unable to identify the file type or even the difference between digital and physical media if we were played music on our own systems. The real magic will always reside at the end of the chain - the loudspeakers.
@Pano, sounds like a good option to me.
@MrKlinky, difference between 16/44 and 320kbps mp3 was obvious to me on NS-10 speakers in a nearfield setup. That said, and as I have written before, sometimes mp3 encoding can be a useful way to clean up a bad recording. The perceptual encoder tends to throw away the non-musical junk and keep most of what is essential. OTOH and IIRC, can't remember her name right now, there is an Italian professor who studies and lectures on perceptual encoding such as mp3. She says her students don't notice anything wrong with mp3 sound when they start her course. However once they learn what to listen for, they can't stand to listen to mp3s anymore.
@MrKlinky, difference between 16/44 and 320kbps mp3 was obvious to me on NS-10 speakers in a nearfield setup. That said, and as I have written before, sometimes mp3 encoding can be a useful way to clean up a bad recording. The perceptual encoder tends to throw away the non-musical junk and keep most of what is essential. OTOH and IIRC, can't remember her name right now, there is an Italian professor who studies and lectures on perceptual encoding such as mp3. She says her students don't notice anything wrong with mp3 sound when they start her course. However once they learn what to listen for, they can't stand to listen to mp3s anymore.
Last edited: