Question about downsampling audio files

[professorwiki]

Hello all,

I am needing to downsample some higher resolution audio files due to playback limitations of the device that will be playing them. The device can downsample itself, but I don't trust it to do a good job. 

So let's say I have a 24-bit, 192 Khz track. I need to convert it to a 24-bit, 96 Khz track. Or the same with a 24-bit, 176.4 Khz track down to 24 bit, 88.2 Khz. What is the best way to get it from one sampling rate to another while minimizing possible digital artifacts? I typically use media monkey to convert files. Is that software capable of doing a good job, or is there another way I should do it? I was reading this article  which talked about one software introducing aliasing and another not, so that made me wonder if I am going about my task the proper way.

Thanks for the help

 

[kimvictor]

  Hello all,

I am needing to downsample some higher resolution audio files due to playback limitations of the device that will be playing them. The device can downsample itself, but I don't trust it to do a good job. 

So let's say I have a 24-bit, 192 Khz track. I need to convert it to a 24-bit, 96 Khz track. Or the same with a 24-bit, 176.4 Khz track down to 24 bit, 88.2 Khz. What is the best way to get it from one sampling rate to another while minimizing possible digital artifacts? I typically use media monkey to convert files. Is that software capable of doing a good job, or is there another way I should do it? I was reading this article  which talked about one software introducing aliasing and another not, so that made me wonder if I am going about my task the proper way.

Thanks for the help

It really shouldn't matter how you downsample them. You're not adding anything to the file or anything. This is not analogue. It's digital. 0/1s don't get added there unless you tell it to. It's like saying that a music copied from SSD sounds better than music copied from HDD, if you say that a software changes a sound(and of course, you don't want it to. Turn off all effects or weird options if they're available).

 

[maeveth]

  Hello all,

I am needing to downsample some higher resolution audio files due to playback limitations of the device that will be playing them. The device can downsample itself, but I don't trust it to do a good job. 

So let's say I have a 24-bit, 192 Khz track. I need to convert it to a 24-bit, 96 Khz track. Or the same with a 24-bit, 176.4 Khz track down to 24 bit, 88.2 Khz. What is the best way to get it from one sampling rate to another while minimizing possible digital artifacts? I typically use media monkey to convert files. Is that software capable of doing a good job, or is there another way I should do it? I was reading this article  which talked about one software introducing aliasing and another not, so that made me wonder if I am going about my task the proper way.

Thanks for the help

There are two basic ways to downsample.  You can drop samples or you can average out the samples over the newly desired sample rate which results in aliasing.  The later will overall produce better sound though your still going to be losing data - no two ways about it.

 

[professorwiki]

  It really shouldn't matter how you downsample them. You're not adding anything to the file or anything. This is not analogue. It's digital. 0/1s don't get added there unless you tell it to. It's like saying that a music copied from SSD sounds better than music copied from HDD, if you say that a software changes a sound(and of course, you don't want it to. Turn off all effects or weird options if they're available).

I think you misunderstood what I meant or I wasn't clear. I'm not *adding* anything or claiming anything analogous to music from a SSD is better than HDD.

What I mean is that if I go from 24-bit 192 Khz down to something like 24-bit 96 or 48, or even 16-bit 48 Khz, that is a lossy process. Similar to how different lossy encoders for formats like Mp3 go about hacking away data differently, I am asking if downsampling from a higher bit depth and/or sampling frequency differs in method and resulting quality depending on the software used. The article I linked suggests that some downsamplers are in fact better than others because some introduce artifacts because of ineffecient downsampling. 

When going from a bigger sampling rate to a lower one, data must be lost as the downsampler must choose which samples to keep and which to exclude. If different software differs in how it goes about deciding which to keep, the quality might be audibly different as an outcome..That is what I meant. Just asking if media monkey does a good job or if I should use something else

 

[RRod]

I have not used media monkey so I cannot vouch for it. The main issue for resampling and quality is if you're having to do it real-time, in which case the default settings of some resampling packages can sound pretty bad. For offline conversion, there are plenty of options for getting a good resample all the way down to even 16/44.1.
 
If you're worried about media monkey, you can try foobar which seems to get much praise. I use SoX personally, and can give you incantations if you want to give it a try.

 

[professorwiki]

That's actually why I'm worried about it. I can't confirm that my android box running kodi is outputting 24/96 over optical. Kodi's settings let me set the max sample rate to 96 Khz, which I think also upsamples lower res files, but android itself I know has a limit of 16-44.1 Khz with dithering and can play 24-bit 44.1 Khz without dithering (but I don't think I need it to dither anyway since it will be fed to a DAC that should take care of that). So, unless Kodi is bypassing android's audio stuff and outputting directly to optical via its own audio drivers, I am assuming everything is being sent out as 16/44 or 24/44. Hence I want to get my files into 24/44 such that no resampling is done on the fly. I'd rather just resample my files as I sync them to an SD card so android doesn't screw with my audio quality, but at the same time want to make sure it's being done well. 

If anyone has a program they have used for downsampling and are happy with the results feel free to let me know. If it's free I can try it and compare to the media monkey's file to see if I can hear a difference

 

[RRod]

  That's actually why I'm worried about it. I can't confirm that my android box running kodi is outputting 24/96 over optical. Kodi's settings let me set the max sample rate to 96 Khz, which I think also upsamples lower res files, but android itself I know has a limit of 16-44.1 Khz with dithering and can play 24-bit 44.1 Khz without dithering (but I don't think I need it to dither anyway since it will be fed to a DAC that should take care of that). So, unless Kodi is bypassing android's audio stuff and outputting directly to optical via its own audio drivers, I am assuming everything is being sent out as 16/44 or 24/44. Hence I want to get my files into 24/44 such that no resampling is done on the fly. I'd rather just resample my files as I sync them to an SD card so android doesn't screw with my audio quality, but at the same time want to make sure it's being done well. 

If anyone has a program they have used for downsampling and are happy with the results feel free to let me know. If it's free I can try it and compare to the media monkey's file to see if I can hear a difference

 
My language was a bit muddy above; I meant that I use SoX for downsampling (non-real-time) and like it, and that many others like foobar for the same task.

 

[professorwiki]

I actually think this topic may be a moot point now. Just received the DAC, and the USB cable it came with works with my Matricom G Box Q. The volume control is disabled, or rather, if I turn it down using the keyboard the audio shuts off. It's either on or off, so I take that as meaning the USB cable is delivering bit-perfect data from the G Box Q? If so, that means I am able to play back the 24/96 and 24/88.1 files without any downsampling occuring

Can anyone confirm I am interpreting this situation correctly?

 

[kimvictor]

Ah, check you audio settings. It shows you what your DAC is capable of playing. You should have an option to change sample rate, which shows you all the rates your DAC is capable of. Check that.
 
What DAC are you using? Cause the same thing happens with my DAC, although it's limited to 16/48. Oh wait, if G Box Q is an android, it's unlikely to do 24bit, unless it specifically says it supports it. Like a lot of androids don't support it, except few companies.

 

[Music Alchemist]

If need be in the future, I would recommend converting your hi-res files to lossless 16-bit / 44.1 kHz with dBpoweramp. Sounds the same as hi-res. You probably wouldn't even be able tell lossless apart from 256 kbps AAC. Of course, you can keep the original files backed up. And if you don't need to convert anything anymore, leaving the files as-is is fine, as long as there is enough hard drive space.

 

[professorwiki]

   
What DAC are you using? Cause the same thing happens with my DAC, although it's limited to 16/48. Oh wait, if G Box Q is an android, it's unlikely to do 24bit, unless it specifically says it supports it. Like a lot of androids don't support it, except few companies.

It's a Maverick Audio Tubemagic D2. It supports up to 24/96. I believe though that using the USB connection for music instead of optical bypasses the android's limitation though. I didn't realize the G-Box Q would support using USB audio, which is why I was previously planning on using optical. With optical I know it does inherit the limitations of android to 16/44, as the fact that you can adjust the volume indicates it is passing through android's sound drivers. But with USB, the volume control does not work, indicating to me that the music is bypassing the sound software controls and being transmitted bit-perfect. That's the impression I've got about using USB for audio on the internet. I've seen people say USB is bit perfect and I've also seen people say things like if software volume controls are disabled then it's bypassing the audio processing in the device, regardless of the OS. In any case, whatever it is, the USB audio connection sounds really nice with the Tubemagic D2. I've got Kodi set to "best match" which plays the files at their encoded resolution. I was playing around with different sound settings, and putting Kodi on the fixed mode and setting it at 96 Khz audio fixed with the highest quality setting for resampling, which sounds good too, but "best match" mode is audibly better. Likewise, setting it fixed at 44.1 made the bass on the higher res files lose fidelity. I was surprised that it was audibly different at all. So, best match mode seems to be ideal, because it just plays the file at the native sample rate and does no resampling for any file. Not sure why putting it at fixed 44.1 makes the bass lose fidelity, but it definitely does. 

 

[kimvictor]

 

It's a Maverick Audio Tubemagic D2. It supports up to 24/96. I believe though that using the USB connection for music instead of optical bypasses the android's limitation though. I didn't realize the G-Box Q would support using USB audio, which is why I was previously planning on using optical. With optical I know it does inherit the limitations of android to 16/44, as the fact that you can adjust the volume indicates it is passing through android's sound drivers. But with USB, the volume control does not work, indicating to me that the music is bypassing the sound software controls and being transmitted bit-perfect. That's the impression I've got about using USB for audio on the internet. I've seen people say USB is bit perfect and I've also seen people say things like if software volume controls are disabled then it's bypassing the audio processing in the device, regardless of the OS. In any case, whatever it is, the USB audio connection sounds really nice with the Tubemagic D2. I've got Kodi set to "best match" which plays the files at their encoded resolution. I was playing around with different sound settings, and putting Kodi on the fixed mode and setting it at 96 Khz audio fixed with the highest quality setting for resampling, which sounds good too, but "best match" mode is audibly better. Likewise, setting it fixed at 44.1 made the bass on the higher res files lose fidelity. I was surprised that it was audibly different at all. So, best match mode seems to be ideal, because it just plays the file at the native sample rate and does no resampling for any file. Not sure why putting it at fixed 44.1 makes the bass lose fidelity, but it definitely does. 

Well, honestly, I'm not so certain. My Galaxy S3 played 24bit music(with a 3rd party app), and played it through OTG out, but I knew that it was being down sampled to 16bit, as the app developer told me that it would, even if I play it through USB. It does kind of suck.

 

[professorwiki]

Hm, too bad the DAC doesn't show what input it is receiving. Still, at least I found the settings which sound best.

 

[goodyfresh]

  Hello all,

I am needing to downsample some higher resolution audio files due to playback limitations of the device that will be playing them. The device can downsample itself, but I don't trust it to do a good job. 

So let's say I have a 24-bit, 192 Khz track. I need to convert it to a 24-bit, 96 Khz track. Or the same with a 24-bit, 176.4 Khz track down to 24 bit, 88.2 Khz. What is the best way to get it from one sampling rate to another while minimizing possible digital artifacts? I typically use media monkey to convert files. Is that software capable of doing a good job, or is there another way I should do it? I was reading this article  which talked about one software introducing aliasing and another not, so that made me wonder if I am going about my task the proper way.

Thanks for the help

If you're only downsampling rather than converting it to lower bit-depth as well, choice of software should not really matter.  If you are planning to also lower the bit-depth, same from 24 to 16, then you may want to look into a program like dbPoweramp which has good options for dithering.

 

[Music Alchemist]

  If you're only downsampling rather than converting it to lower bitrate as well, choice of software should not really matter.  If you are planning to also lower the bitrate, same from 24 to 16, then you may want to look into a program like dbPoweramp which has good options for dithering.

 
Downsampling lowers the bit rate (kbps) automatically. You're probably thinking of bit depth. (24-bit, 16-bit, etc.)