Why Use High Sample Rates?

This is a question I see often: if the final form of the project is 44.1kHz or 48khz, why bother recording at high sample rates? There isn’t any one answer to that question, there are lots of answers. But I thought it might be useful to illustrate a reason by using a real world example…. And if you cannot be bothered reading all of this, then just remember this: Plugins LOVE dense data!

192k

When I was working on a film last year called HOME BY CHRISTMAS I had the challenge of creating the sounds of memories – sometimes beautiful, happy romantic memories and other times the horrific memories of war, World War II to be exact. The film deals with the directors father telling stories about his experiences in WWII, first in the infantry and then as a prisoner of war. Before I started work I had a number of discussions with the director about how to approach the archival footage. One sequence in particular showed a distant battle at night with flashes of bombs going off and we realise this is memory and not ‘reality.’ So we didn’t want any full resolution dynamic explosions, and the scene ends with an introspective moment with her father in the garden of his house now, thinking about it all. So amongst other things I went looking for sounds that could evoke falling bombs and one of the elements I ended up using was a screech from a falling fireworks, except I processed the hell out of it. To start off with I pitched it down, way down. I sent it from SoundMiner into ProTools at 17% real speed, so what was a high fast screech became a long drawn out doppler that was almost vocal. But the problem then became the grain – the original recording was recorded on DAT quite a few years ago so was only 16bit 48k and pitching that much revealed the grain nastily. Partly to hide the grain but also to push the sound into memory I started playing around with processing it through various impulse responses – some were natural acoustics while others were weirder IRs (such as the Sanitarium collection) – with a little help in the final mix from the lovely reverbs at Park Road Post, the end result worked beautifully! But it made me think: I need some source material like that recorded at high resolution!!! And so the idea of recording the FIREWORKS library at 192kHz became a fixture in my mind.

Now the annoying thing is I don’t actually own a 192kHz interface, my ProTools HD2 rig has a 96io interface so I cant actually play or edit sounds at 192kHz. Back when i recorded the SEAL VOCALS library I actually recorded it at 192kHz but had no choice but to edit it at 96k, but this time I was determined to pursue the 192kHz from start to finish. So I called a favour in from a friend and rented his 192io for a week and set to and edited all the FIREWORKS material as well as recutting the SEAL VOCALS library (which will be a free update next week to anyone who bought SEAL VOCALS MAX)

Anyway once I had finished editing & output the FIREWORKS 192k library I started experimenting with plugins and tried some extreme pitch shifting. Using the plain old Digidesign Pitch AudioSuite plugin I took a sound with plenty of high frequency content and pitched it one octave down. Beautiful!! Totally continuous with no grain artefacts. Ok lets try 2 octaves… Wow!!! I was sold! Here is that first trial (a Predator firework) first at normal speed and then at -24 semitones down.

Fireworks 192k pitchshift by timprebble

Heres another example (a Red Dragon firework) first at real speed, then one octave down, then two octaves down:

Fireworks 192k pitchshift 2 by timprebble

Heres a more percussive sound (a Solar Flare firework) – first at real speed, then 1 octave down, then 2 octaves down…

Fireworks 192k pitchshift 3 by timprebble

And a Might Cannon firework fired from inside a pipe, first at real speed, then one octave down, then two octaves down

Fireworks 192k pitchshift 4 by timprebble

Now obviously these examples are no longer full rez – the Soundcloud embed is limited to 128k MP3, if you go to the Soundcloud page they sound a bit better, but believe you me they do sound great!!! Apart from on their own such deeply altered sounds can also really contribute when layered with real sounds eg imagine a slow motion body fall and layering one of those Mighty Cannon hits pitched down two octaves! With real speed sounds layered on top your ear wont read the pitched layers as being altered due to masking, but the ooomph of thaose slwoed sounds is potentially VERY useful!

So apart from buying a 192io interface or one of the new Apogee Symphony interfaces, what use is all of this to you? Well there is a quick & dirty way to alter the playback speed of high resolution files in ProTools. If you go to the Workspace Browser in ProTools and find eg a 192kHz file, if you scroll across to the sample rate you will find it can be edited! So if I take a 192khz file and edit its sample rate to be 96kHz it will then play at half speed (one octave pitch shift) when it is imported into a 96kHz session! Now I haven’t compared this method with pitch shifting a sound in a 192kHz session and then exporting it to 96khz, but I will!

But as I mentioned at the head of this rant, the real moral of the story is that plugins LOVE dense data. Whether it is linear or variable pitch shifting or time stretching, the denser the data is in the source file, the more data the plugin has to use when interpolating. So even if your work session is 48kHz, if you intend to do some serious pitch manipulation it would be well worth your while recording the source files at 96khz or 192kHz and doing that processing in a 96khz session. Then transfer the processed files back into the 48k session for syncing/editing etc…

26 Responses to Why Use High Sample Rates?

  1. Pingback: Tweets that mention Music of Sound » Why Use High Sample Rates? -- Topsy.com

  2. Pingback: Interview, Workspace Tour: Alessandro Cortini’s SONOIO Album and Synth, Why Use High Sample Rates?, Review: Korg Monotron, The definitive Beastie Boys sample source collection, Good-looking Ikeahacked speakers, Makunouchi Bento - Swimé

  3. Davide favargiotti says:

    Great post!
    Changing the sample rate in protools workspace should be like import audio without applying SRC (in the import audio menu).
    Happy pitch shifting!

  4. Wit says:

    Nice post! the quote “Plugins LOVE dense data” explains this post very well.

  5. Pingback: Music of Sound » Why Use High Sample Rates? | Box of Toys

  6. Ian Palmer says:

    Man, pitching the fireworks down 2 octaves really does sound amazing!!! Particularly like the Red Dragon one, now to convince my boss to fork out!

    Nice one Tim!

  7. Dan says:

    It’s also worth pointing out that during the mixing process, higher sampling rates reduce the distortions introduced by digital EQ, compressors, and most other processes, including the summing itself. So by doing all your processing/mixing at a high resolution, your final mix is as best as it can be, which can then be dropped to your release rate as a single file (ie. without compounding the errors).

  8. Steve Urban says:

    Beautiful post Tim. Even at lo rez, the examples illustrate your point wonderfully.

  9. Pingback: Designing Sound » FIREWORKS, New Multi-Channel SFX Library of HISS and a ROAR [with Exclusive Stories]

  10. Tuomas says:

    I hear not much high frequency content on those -2 octaves files, which of course only makes sense. I wonder what the difference would have been if you would have done the same with files recorded originally at 48 kHz? Or with files recorded at 196, converted to 48 kHz, and then re-converted to 196 when imported into 196 kHz session?

    • tim says:

      In answer to your first question, are you telling me you have never tried pitch shifting a sound down 2 octaves at 48kHz? As I said in the post have plenty of times… Why don’t you try it and make your own conclusions? Not sure I understand the second question

  11. Tuomas says:

    Have done that many times of course. Second question was that wouldn’t sound recording at 196 kHz – pitched down 2 octaves sound same than the same sound recording at 196 kHz – converted to 48 kHz – re-converted back to 196 kHz – pitched down 2 octaves, IF the original recording doesn’t have any info above 20 kHz (which is rarely the case)?

    • tim says:

      It would be an interesting test to do, both digitally and analogue… It would depend on the processes involved, ie if digitally, dithering down to 48k and then interpolating back up… analogue would depend on quality of convertors in each direction… There would have to be degradation but who knows how noticeable it is…

      I’ll contact a local DSP engineer and ask their opinion…

      • max says:

        I would also be very interested in how big the difference would be if the source material was at a sample rate of 48 kHz instead of 192 kHz. Theoretically there should not be a big difference if there’s not much going on above 20kHz (and if there is, how much does it really matter when pitch shifting), but who knows what really happens in the process…

    • JJ Benson says:

      My mics will record useful information at 30kHz+ even though they are rated 20-20Khz. There is a drop off in sensitivity, however. With Hi Rez recording that extra range would be useful compared to the upper limit of 24kHz recording at 48kHz.

  12. Rene says:

    with regards to workflow – this is why I swear by soundminer. Even if my work session is at 48k, SM will let me audition the sounds at native resolution, then varispeed them down to whatever sounds right,then spot them in at the new speed and 48k rez. Superfast and easy way to play with high samplerate sounds while still in the creative mindset.

  13. JJ Benson says:

    Great examples, can already see using the Red Dragon on a crashing fighter jet…

  14. Pingback: Recording at High Sample Rates | Colin Hart's Sound Kitchen

  15. max says:

    I really LOVE the reflections off the mountain in the recordings. You can create things like this with Altiverb as well and it works wonderfully, but in my experience there’s always a limit to it because you start to hear the speaker (or whatever it is). So maybe it would even be a good idea to use an extra stereo-set of mics set up to only capture those wide reflections from the mountains, without the direct sound…. maybe you guys did that, not sure, there ARE a lot of channels in those multitrack-recorings :)

    • tim says:

      aye, the pair of 816s that Ray used in his quad setup were mostly pointed at the slap delay off the hills…. that slap is beautiful & very difficult to recreate artifically!

  16. Pingback: 192k Experiments

  17. Pingback: The Art of Sound Design with Tim Prebble | Schurr Sound

  18. Juhani says:

    Great info, very inspiring! I also have come to the conclusion that the highter the sample rate, the better the time stretching sounds. It´s almost like magic…

    A tip (that I´m sure you already know of): to listen to things at half speed (ie half sample rate, I believe) in Pro Tools, just press Shift Spacebar…

    /Juhani

    • tim says:

      ah yes, been using shift spacebar for a decade or more…. ditto for CNTRL numeric keypad for playback at 1/2, 1/4 and 1/8 speed, backwards & forwards…..

  19. Pingback: Digital Audio and Digital Image | katepowerpr

  20. Jake says:

    Hi Tim,

    Cool to hear your experiments. i’m interested in exploring really slowed down audio as well for a project. Can you recommend some devices for recording at extreme rates, for then slowing down audio? Ie the most powerful recorders for this purpose? and the most value for money?

    I also want to synch it to video eventually but i’m not too worried about that as surely that’s gonna be failry straight forward to figure out.

Leave a Reply

Please use your real name instead of you company name or keyword spam.