Upsampling on the fly, the future of digital music?

[WAD62]

Hi all,

This is inspired by Gaz & Ali's squeezebox touch enhancements thread, but it's probably worth a thread on its own as a more generic topic.

I've been playing around with SoX (Sound Exchange), a plugin for many computer music players, and is available for SqueezeBox.

http://sox.sourceforge.net/

My squeezebox setup is not appropriate for upsampling, I only have a QNAP TS-119 (1.2ghz) as a server, and from that I'm running 2 receivers and a SB3. So the server isn't meaty enough to upsample for 3 clients, and the clients are limited to 48k anyway.

However I also run an instance of winamp against the same FLAC library I use for squeezebox, from my desktop configuration...XP/winamp/m-audio 2496/Mini-T/mission 780SEs(upgraded)

I've installed the FFSoX plugin for winamp, it is an input plugin and can be used to upsample, and also apply replay gain processing, prior to winamp playing it.

http://in-ffsox.sourceforge.net/

There are several parameters one can apply, for re/upsampling, the sample depth (from 16 to 24 bit) and the frequency (up to 192kHz), with a upsampling quality option (set this to very high, a bit of a CPU chomper), along with some dither parameters, and ReplayGain processing (including a pre amp, very useful if you use ReplayGain!!!).

Apparently handling ReplayGain in this manner, i.e. prior to playback, has advantages too.

I've set the upsampling to the highest my soundcard will handle, 24/96, applied my replay gain in SoX ('Album' with a +3 preamp setting). So as I look at the winamp playback screen I'm getting 4608KBPS @ 96kHz (as opposed to 1411KBPS @ 44.1kHz).

Put quite simply the sound quality is better, clearer smoother mid and top end, more control, the sort of improvement you'd hear from a better quality/more expensive source. I doubt what I'm hearing would compare to a full HD version of the recording, but it is a noticable improvement on the original CD, which sounds identical to normal FLAC playback (no surprise there).

I have a lot of money invested in 'red book' cd's and am not about to re-buy everything at high res, when and if it becomes available. These algorithms will improve, along with processing capabilities, which means that my standard FLAC library will have the capacity to sound as good as the upsampler being used.

This has to be the way to go, store and stream the music as standard compressed FLAC, and then upsample at playback time. One alternative would be to convert the source, which would take quadruple the storage, and also the transmission bandwidth.

The other is to upsample at the server, but this would require a very beefy server (scalability would be an issue), and one would still have the bandwidth issues, with multiple players.

Someone needs to come out with a network player with SoX capabilities at the client end...currently it's called a PC!!!! ;)

Get your fingers out Logitech, I want a new receiver with a 1Gb processor, on board SoX, and a digital out...shouldn't be too difficult :doh:

[Gazjam]

Hey Will,
just buy a Touch, you know you want to!

Agreed on the benefits of upsampling..I've sanity checked it a few times comparing 44khz to upsampled 96khz..sounds more "solid" and clearer...music seems to flow better.

Subjective? Mibbay...
If you dont at least TRY it you'll never know.

I dont use Replaygain Will, so no idea on the effects of either upsampling or Soundchecks Mods on it..but I cant see the harm in it, considering its just a matter of increasing/decreasing the gain.

Gaz

[WAD62]

Hey Will,
just buy a Touch, you know you want to!

Agreed on the benefits of upsampling..I've sanity checked it a few times comparing 44khz to upsampled 96khz..sounds more "solid" and clearer...music seems to flow better.

Subjective? Mibbay...
If you dont at least TRY it you'll never know.

I dont use Replaygain Will, so no idea on the effects of either upsampling or Soundchecks Mods on it..but I cant see the harm in it, considering its just a matter of increasing/decreasing the gain.

Gaz

I presume the SoX algorithm used by squeezebox must be the same as the winamp bolt on.

Do you have options for upsampling quality, dither, noise shaping and all that malarkey?

I'm glad you hear an improvement too, with similar characteristics, it's not just me going mad, it is definitely better to my ears. Can you go to 192k with the touch? My soundcard maxes out at 24/96.

I can see the big advantage of upsampling, and replay gain application, before it hits the player, as the player no longer needs to do any DSP processing, leaving it just to feed the DAC.

I'm afraid the Touch wouldn't help me, my issue is that I'd need to replace my server and all 3 clients to do it using the current available products, and I don't think doing it at the server end is the best long term solution, particularly with a variable number of clients.

I need the invention of a new product, a beefed up receiver with on board SoX processing, and a 192k DAC...or I use a PC in the meantime. ;)

Anyway my main DAC (Audiolab 8000DAX) is only 48k capable, and I don't think that there's any point in upsampling to 24/48, so I'm thinking about the future. A nice compact FLAC library, streamlined transmission of FLAC, and a beefy client with upsampling on board, preferably controllable by squeezecenter...I want the moon on a stick! :eyebrows:

[Reid Malenfant]

Hey, is this thread restricted to upscaling streamed audio or anything on the fly? :scratch:

If open to all here is my take on it, installing shortly ;)

3438

[WAD62]

Hey, is this thread restricted to upscaling streamed audio or anything on the fly? :scratch:

If open to all here is my take on it, installing shortly ;)

3438

Hi Mark,

It's more on the general concept of upsampling streamed audio, but I imagine that the upscaling of video is a suimilar concept too, if not more complex!!!

If I can see the improvements made by a decent upscaler for video, then why not an upsampler for audio? It's probably a far less complex goal to achieve, particularly if it occurs before hitting the 'player'...;)

[Reid Malenfant]

Will, i posted a pic of a dCS Purcell (upsampler) & Delius DAC, these are purely for audio my friend.

Transport > Purcell (upsamples to 24bit 88.2, 96, 176.4 or 192) > Delius 24 bit DAC = :cool:

[WAD62]

Will, i posted a pic of a dCS Purcell (upsampler) & Delius DAC, these are purely for audio my friend.

Transport > Purcell (upsamples to 24bit 88.2, 96, 176.4 or 192) > Delius 24 bit DAC = :cool:

Ah but you said 'upscale'...:ner::ner: in pedant mode ;)

Perhaps the word 'audio' should have been a clue :doh:

Exactly what SoX is doing on my PC, I think, i.e. upsampling before it hits the Player/DAC, must be following similar principles I imagine, and with a similar effect :)

I presume you demoed the beast before purchase, what improvements did you notice? Similar to myself and Gaz's observations?

Oh and do you find an additional improvement between 96 & 192 ? As I don't have a soundcard capable of 192

[Reid Malenfant]

Ah but you said 'upscale'...:ner::ner: in pedant mode ;)

:doh: So i did, slightly silly of me eh? :scratch:

If you have ever gotten a listen to an SACD in a good system then you can possibly imagine how the dCS kit sounds :)

It does all that you mentioned, control of dither, noise, output filters with loads of settings for each :eek:

There is a subtle difference between 96 & 192 though i'll be using the DSD (direct stream digital) connection via firewire as this is the dogs danglies & should result in even better sound. I even have 4 different firewire cables here to experiment with :eyebrows:

[WAD62]

:doh: So i did, slightly silly of me eh? :scratch:

If you have ever gotten a listen to an SACD in a good system then you can possibly imagine how the dCS kit sounds :)

It does all that you mentioned, control of dither, noise, output filters with loads of settings for each :eek:

There is a subtle difference between 96 & 192 though i'll be using the DSD (direct stream digital) connection via firewire as this is the dogs danglies & should result in even better sound. I even have 4 different firewire cables here to experiment with :eyebrows:

Ok Mark but presumably you're upsampling before you send out on the firewire? :scratch: or are you just sending 16/44.1 via the DSD?

[Reid Malenfant]

Ok Mark but presumably you're upsampling before you send out on the firewire? :scratch: or are you just sending 16/44.1 via the DSD?
Will, i'm not that up to date on how it all works in all honesty :eyebrows: Each manual is the thickness of a paperback, all i know is that it does what it says on the tin :lol:

As far as i know DSD is already something other than standard 44.1KHz digital as it appears to be some kind of 2.6MHz (or there abouts) signal :scratch:

[WAD62]

Will, i'm not that up to date on how it all works in all honesty :eyebrows: Each manual is the thickness of a paperback, all i know is that it does what it says on the tin :lol:

As far as i know DSD is already something other than standard 44.1KHz digital as it appears to be some kind of 2.6MHz (or there abouts) signal :scratch:

That's a big tin with a lot of writing on it then...:lolsign:

Looks like massive upsampling from what I can glean (64x!!!!), err I'm running 2.5x...just spotted some of the prices, no wonder your Scottish mate wasn't bothered by the price of his server!!! :stalks:

I'm curious as to how close an upsampler can get you to an SACD version of of a standard CD, from the basic CD (or FLAC file)...certainly gives more mileage to ones existing 16/44.1 collection....;)

Edit; One little problemette of using SoX to upscale for winamp is that FF and REW are no longer functional, makes sense really as this is an input plugin, and winamp has no control over it at run time, it is merely the recipient of the upscaled data.

[technobear]

As an engineer, the idea of upsampling 44.1 kHz to 96 kHz or 192 kHz fills me with horror. In order to perform this operation, the software has to compute entirely new values for each and every sample. The final datastream has nothing whatsoever in common with the starting data.

It seems however that the benefits of being able to use more benign digital filtering outweigh the loss caused by the horrors of translating the datastream into something else entirely.

It seems to me that the best way to take advantage of upsampling is to use a sample rate which is an integer multiple of 44.1 kHz, e.g. 88.2 kHz or 176.4 kHz. In this scenario the original data is still present in the output datastream. This ought surely to give a more accurate rendition of the piece.

The main reason manufacturers went for 96 kHz or 192 kHz is because the chips to operate at these sample rates were available very cheaply for use in DVD players.

Thankfully there are a number of units on the market now that will perform synchronous upsampling to 88.2 kHz or 176.4 kHz and then convert to analogue at those frequencies.

[WAD62]

As an engineer, the idea of upsampling 44.1 kHz to 96 kHz or 192 kHz fills me with horror. In order to perform this operation, the software has to compute entirely new values for each and every sample. The final datastream has nothing whatsoever in common with the starting data.

It seems however that the benefits of being able to use more benign digital filtering outweigh the loss caused by the horrors of translating the datastream into something else entirely.

It seems to me that the best way to take advantage of upsampling is to use a sample rate which is an integer multiple of 44.1 kHz, e.g. 88.2 kHz or 176.4 kHz. In this scenario the original data is still present in the output datastream. This ought surely to give a more accurate rendition of the piece.

The main reason manufacturers went for 96 kHz or 192 kHz is because the chips to operate at these sample rates were available very cheaply for use in DVD players.

Thankfully there are a number of units on the market now that will perform synchronous upsampling to 88.2 kHz or 176.4 kHz and then convert to analogue at those frequencies.

I know what you mean, any messing around with the source worries me.

But one hears what one hears, and as it's non destructive it's worth a play.

There also seem to be better or worse upsampling algorithms, some that just echo the previous sample as it were, and others that try to create a mid point between the previous and the next.

SoX with winamp is doing as you state, upsampling before playing (via winamp), and therefore way before conversion.

[magiccarpetride]

I know what you mean, any messing around with the source worries me.

But one hears what one hears, and as it's non destructive it's worth a play.

There also seem to be better or worse upsampling algorithms, some that just echo the previous sample as it were, and others that try to create a mid point between the previous and the next.

SoX with winamp is doing as you state, upsampling before playing (via winamp), and therefore way before conversion.

My completely unfounded, indefensible hunch is that if a DAC is given a denser digital content, it will be less of a hassle for it to work on converting it to analog signal. The reality is that a 16 bit/44 kHz signal is fairly sparse, and it is left to the DAC to do a lot of heavy lifting in real time in order to reconstruct an analog curve from the very choppy digital approximation.

As soon as we feed a denser (i.e. 24 bit/96 kHz) signal to the DAC, it gets easier to reconstruct the analog curve.

This why, I believe, we can hear less strenuous sound coming out of a higher definition digital signal.

The only thing I'm not sure about is whether 'faking it' (i.e. pretending that a mere 16 bit/44 kHz signal is actually a 24 bit/96 kHz signal by second-guessing and interpolating a lot of info into it before sending it off to a DAC for conversion) can indeed fool the ear.

Discuss...

[WAD62]

My completely unfounded, indefensible hunch is that if a DAC is given a denser digital content, it will be less of a hassle for it to work on converting it to analog signal. The reality is that a 16 bit/44 kHz signal is fairly sparse, and it is left to the DAC to do a lot of heavy lifting in real time in order to reconstruct an analog curve from the very choppy digital approximation.

As soon as we feed a denser (i.e. 24 bit/96 kHz) signal to the DAC, it gets easier to reconstruct the analog curve.

This why, I believe, we can hear less strenuous sound coming out of a higher definition digital signal.

The only thing I'm not sure about is whether 'faking it' (i.e. pretending that a mere 16 bit/44 kHz signal is actually a 24 bit/96 kHz signal by second-guessing and interpolating a lot of info into it before sending it off to a DAC for conversion) can indeed fool the ear.

Discuss...

I concur with your hunch about the DAC's reduced workload, but I don't know enough about their inner workings for it to be any more than just that...

With regards to 'faking it', like most things I think upscaling will improve, but will it reach some kind of ceiling?

If it were a simple graph curve then no problem, but I'm sure the task of getting back to a contiguous soundwave is probably a damned site tougher. :scratch:

There are more questions than answers...

[serendipitydawg]

Thanks for posting this Will. I've downloaded and installed a SoX plugin for foobar and I have to say it sounds wonderful.

The sceptic in me says "You can't do this... bits is bits etc etc" but if it's free and the software works, let's give it a go.

Incidentally my PC didn't like Winamp or the SoX player.

I am sufficiently interested in upsampling now to consider building a silent PC purely to do this. Does anyone think/know if an Intel Atom processor has enough processing power to do this? I am thinking of this purely because you can get away without a processor fan with the Atom.

Any other suggestions welcome.

Cheers!

[WAD62]

Thanks for posting this Will. I've downloaded and installed a SoX plugin for foobar and I have to say it sounds wonderful.

The sceptic in me says "You can't do this... bits is bits etc etc" but if it's free and the software works, let's give it a go.

Incidentally my PC didn't like Winamp or the SoX player.

I am sufficiently interested in upsampling now to consider building a silent PC purely to do this. Does anyone think/know if an Intel Atom processor has enough processing power to do this? I am thinking of this purely because you can get away without a processor fan with the Atom.

Any other suggestions welcome.

Cheers!

It does sound nice doesn't it, I was sceptical too :)

At this moment in time the 1.6Ghz Atom would be the best option IMHO, perhaps even the ASUS Revo. It should have more than enough grunt to upsample 1 stream, even to 192.

I doubt you'd need more than 1Ghz if it was a dedicated machine.

I think for the future an Android based epad would be ideal, but as yet there aren't any players offering this functionality, but they are being developed, there's only a basic winamp beta, and a SqueezeBox player is also in development

Strange about Winamp, I've not experienced any trouble with it before, where did you get the problem?

[WAD62]

...having said that about Winamp, I've just spent an unsuccessful day trying to get the latest version of winamp to play properly via ASIO on the missus's laptop, as she has a USB DAC.

So winamp is fine for an on board soundcard ((I use an M-Audio 2496), but not so good for USB DACs.

I might try Foobar or MediaMonkey for her machine.

[serendipitydawg]

Exactly my scenario Will. Trying to get Winamp to work with ASIO and a USB DAC. Foobar may be the way to go.

As my PC is Windows XP, ASIO is the best sounding optio. The route is foobar asio out dll via USB to mAudio Transit (which comes with it's own ASIO driver) and Toslink to my Beresford TC7510.

[WAD62]

Exactly my scenario Will. Trying to get Winamp to work with ASIO and a USB DAC. Foobar may be the way to go.

As my PC is Windows XP, ASIO is the best sounding optio. The route is foobar asio out dll via USB to mAudio Transit (which comes with it's own ASIO driver) and Toslink to my Beresford TC7510.

Looks like support for the winamp ASIO plugin died about 2 years ago, strange...

[WAD62]

FYI I've managed to get winamp to work with ASIO out to my M-Audio 2496 soundcard, playing FLAC (not mp3), gapless at 24/96.

I used the Japanese winamp output plugin (.dll) referenced on this site, the one on the winamp site does not seem to be supported/work any more...

http://www.aqvox.de/Asio-USB-Audio-installation-e.htm

...just follow the instructions, set the thread to highest priority, select gapless, put the buffer size at max 63, leave the rest as is, apart from pointing at the M-Audio ASIO drivers. N.B. You might need to change the shift output channels from 0 (stereo L & R) to the channel that corresponds to SPDIF, my guess would be 3...

The only other thing is to amend the buffer size via the M-Audio control panel, I've set it to the max of 2048 samples, necessary for gapless and stutter free on my machine, but it is an old one. ;)

There's a marginal improvement for me by a gnats arse (the larger explicit buffer sizes probably help), and I loose crossfade, but I'll give it a go for a while.

[nat8808]

I've set the upsampling to the highest my soundcard will handle, 24/96, applied my replay gain in SoX ('Album' with a +3 preamp setting). So as I look at the winamp playback screen I'm getting 4608KBPS @ 96kHz (as opposed to 1411KBPS @ 44.1kHz).



Those are simply the bitrates (not byte rates) of standard 24/96 and 16/44.1 you should know - although they've used 'Kilo' as 1000 rather than the 1024 it really should be in binary computer terms (fairly common incorrect usage that can make comparison confusing to those who don't realise) - so that is only confirming Winamp is actually playing uncompressed format.

With regard to upsampling, it has always been considered in studio terms (and perhaps hifi techy circles?) that changing bit-rate or sample-rate using algorithms in software on do-it-all processors is second-rate compared to dedicated hardware/chips designed specifically to do the job.

There are plenty of current hifi components and even DAC chips or receiver themselves that upsample etc in hardware so that getting the computer to do it isn't really necessary apart from the fact that some software is free so it's good for just experimenting.. No need to stress out your processing power!

The whole upsampling etc thing purely for better sound came to the hifi scene about 11 years ago so is well established by now.

The Berhinger SRC2496 is supposed to be pretty good, especially at £140 new - so must be very cheap on Ebay by now. Or I've seen dCS Purcells sell for as little as £400 without firewire.

[WAD62]

Those are simply the bitrates (not byte rates) of standard 24/96 and 16/44.1 you should know - although they've used 'Kilo' as 1000 rather than the 1024 it really should be in binary computer terms (fairly common incorrect usage that can make comparison confusing to those who don't realise) - so that is only confirming Winamp is actually playing uncompressed format.

With regard to upsampling, it has always been considered in studio terms (and perhaps hifi techy circles?) that changing bit-rate or sample-rate using algorithms in software on do-it-all processors is second-rate compared to dedicated hardware/chips designed specifically to do the job.

There are plenty of current hifi components and even DAC chips or receiver themselves that upsample etc in hardware so that getting the computer to do it isn't really necessary apart from the fact that some software is free so it's good for just experimenting.. No need to stress out your processing power!

The whole upsampling etc thing purely for better sound came to the hifi scene about 11 years ago so is well established by now.

The Berhinger SRC2496 is supposed to be pretty good, especially at £140 new - so must be very cheap on Ebay by now. Or I've seen dCS Purcells sell for as little as £400 without firewire.

At what point did I mention the word 'Byte', after more than 25 years in IT I'm more than aware of the difference...and a nibble is? ;)

KBPS is simply the terminology winamp uses, and is useful to identify the source it is playing, and hence that upsampling is actually occuring :doh:

This thread was about the ability to do it for free on a laptop, with the latest software (technology moves on at a pace), not about buying an expensive legacy dedicated appliance :doh:

Perhaps I should have been more specific in the title...

[nat8808]

Don't keep hitting your forehead like that - you'll flatten it.

I never did comment or imply that you were making a mistake in calling it bytes or bits - I simply put that bit in brackets to emphasize what I was saying. Funny how forums don't work well in this way - people often take things as criticism.

Your question in the title I took to mean that you'd only just discovered it and thought it to be a revolution whereas the 'future of digital' has been around for years and years and a lot of recording software and free software that can do this has been around for a long time. I carried that sentiment through to reading your line about what Winamp was displaying, especially as you were expressing uncompressed audio in terms of a bitrate which is (or is to me) unusal as it's normally reserved for use in comparing compressed formats. Someone only used to compressed formats might think this was an amazing bitrate as opposed to just normal for uncompressed - so I was letting you know if that was the case.

As I said, that some software for doing upsampling can be found for free is a good reason to try it out and so was giving you the benefit of doubt that this is what the thread is ostensibly about. You may also be trying to do this completely from within the laptop using its own line or headphone out.

That said, there are many dac chips and receiver chips out there for a few pounds that will upsample on the fly too using dedicated, especially designed silicon. There are many DACs using these chips - if you want good sound from your laptop then surely you'll be using an external DAC? Therefore there won't be additional expenditure if you buy the right dac and so neither will you have to use up CPU time performing the same function.

Also, if you did want dedicated external boxes to do it then they are now cheap, not expensive at all yet good software that is comparable in quality to a good hardware unit will be expensive.

[Vincent Kars]

There are indeed dedicated DSP chips available.
Do they perform better than software?

One thing is obvious, an algorithm is a algorithm.
You have to program it anyway.
We might wonder if the same algorithm programmed in software or in hardware will yield different results.
DSP chip comes in all kind of capabilities.
Performance and precision varies.
PC floating point is internally done in 128 bits
Best DSP chips have 48 bits + a 76 bit accumulator.
Cheaper DSP chips have less so you might expect a higher quantization error.
I wonder if we can conclude that a hard ware solution is a priory superior to a software one.

[WAD62]

Don't keep hitting your forehead like that - you'll flatten it.

I never did comment or imply that you were making a mistake in calling it bytes or bits - I simply put that bit in brackets to emphasize what I was saying. Funny how forums don't work well in this way - people often take things as criticism.

Your question in the title I took to mean that you'd only just discovered it and thought it to be a revolution whereas the 'future of digital' has been around for years and years and a lot of recording software and free software that can do this has been around for a long time. I carried that sentiment through to reading your line about what Winamp was displaying, especially as you were expressing uncompressed audio in terms of a bitrate which is (or is to me) unusal as it's normally reserved for use in comparing compressed formats. Someone only used to compressed formats might think this was an amazing bitrate as opposed to just normal for uncompressed - so I was letting you know if that was the case.


Too late it's already flat...:)

Non intended as they say...

It wasn't the clearest thread title I must admit, unfortunately one can't change it once it's done.

[WAD62]

There are indeed dedicated DSP chips available.
Do they perform better than software?

One thing is obvious, an algorithm is a algorithm.
You have to program it anyway.
We might wonder if the same algorithm programmed in software or in hardware will yield different results.
DSP chip comes in all kind of capabilities.
Performance and precision varies.
PC floating point is internally done in 128 bits
Best DSP chips have 48 bits + a 76 bit accumulator.
Cheaper DSP chips have less so you might expect a higher quantization error.
I wonder if we can conclude that a hard ware solution is a priory superior to a software one.

Hi Vincent,

As you say in both instances it's a software algorithm running on some form of CPU, will we soon be at a point where dedicated up samplers become obsolete, in the same way as for example dedicated CD copiers have.

Particularly when one considers that the processing power of players will only increase, ipads and epads might make excellent players for the future...

[realysm42]

If I have my PC set to output sound at 24bit 192Hz and the same with my sound card, but Foobar is playing at 44Hz and without "Super mode" (whatever that is?!) does this mean it's all for nothing?

Do I need to tell Foobar to play at 96Hz (with super mode on) to get the best results?

If yes, why is Foobar bottlenecking what my hardware is capable of?

Also, whch players would allow me to use my hardware's full capacity please?

[Welder]

I don't understand what you are trying to achieve Martin or what equipment you're trying to achieve it on.

Are the files you are trying to play 24/192?
If they are not, why are you trying to play them at 24/192?

[bobbasrah]

Interesting outcomes indeed. I liked the hunch that the wave generation was smoother, as reasonig why the following stages might sound better, entirely plausible. At least it is a better attempt at explanation than voodoo/synergy.
Hirez music is the future because the industry can flog it, this methodology is a bridge to it for lower resolution formats.
Digital TV took 100 channels of analogue crap and now we enjoy.....yes, well....

[realysm42]

I don't understand what you are trying to achieve Martin or what equipment you're trying to achieve it on.

Are the files you are trying to play 24/192?
If they are not, why are you trying to play them at 24/192?

All of my files are 44Hz/16 bit.

I guess I'm asking what's the point in having these options on some items (ie my hardware) only to be bottlenecked by the player, at the end of the line.

It's more of a hypothetical question really.

[Vincent Kars]

It is by and large a matter of configuring your system.
If you set Foobar to use DS and Win to 24/192, all audio will be upsampled by win to 24/192.
If you use e.g. ASIO and set your sound card to 24/192, the sound card will take care of the upsampling.
If you combine Foobar with Sox, Sox will do the upsampling.
Etc.

[realysm42]

Nice one :yesbruv: