Hey ya'll,

I have questions about sample rates for audio used in video.

Audio cds 44100 but video/tv/movie soundtracks are 48000. So what happens to cd quality audio when added to video - is it "up sampled" somehow? Does it change it for better or worse? Would it be better to import the soundtrack at 48000 to begin with? Is it something to even be concerned about?

In the past, I've just put the mixed down "cd" 44100 version on the few youtube videos I've done and never thought about it. But now that I'm aware of this - which actually I don't know about youtube specifically - but I got a new video editor and I saw that the sound is rendered to 48000 when the video is rendered. I tried searching on the subject but I'm really not clear....

Thanks in advance for your help and knowledge.

Josie

Hi Josie,

The 44,100 Hz and 48,000 Hz is the rate at which an audio signal is processed in relation to the signal's variation with time. Below is a simplified image of a sound wave.



Using the above image ...

1. The sound starts at where I've shown.

2. One second later it has traveled up to a maximum value, gone back down past the horizontal axis to a minimum value and then returned back to the horizontal starting line. (I've marked this movement by the red double-ended arrow. It's this horizontal variation in sound that we hear as pitch; the vertical variation we hear as volume.)

3. If a sample rate is 44,100 Hz, the above one-second variation in movement of this sound wave is captured as 44,100 individual snippets of information.

4. If the sample rate is 48,000 Hz, then the variation of the sound wave's movement during one second is captured as 48,000 individual snippets of information.


In going from one sampling frequency to another one, the best method is to use the original data to calculate new values for new rate. There are software programs to do this and it might be that your video software has this ability.

As a consequence of my net searches, I've just stumbled across this ...

link to "Music Editor", a freeware re-sampling program. (I haven't tried it so I have no idea how effective it is.)

Thanks for asking this question! It's inspired me to find out some stuff.

All the best,
Noel

I've been studying the third edition of Bobby Owsinski's "The Mixing Engineer's Handbook". In addition to many other details he mentions preparing a song for iTunes.

Songs distributed on iTunes are encoded for distribution by Apple using an exclusive encoding scheme they developed. The iTunes store has an app mastering engineers can use to test and verify a song is "iTunes Ready" prior to submitting a song.

In addition to other requirements, the audio file should be rendered at 24 bits and 96kHz. Apple says the intent of this requirement is to future proof the high fidelity quality of the audio file as encoding schemes change in the future.

Submitted files that meet the requirements and pass the app test are identified with an iTunes ready icon.

Hey Noel.

You always go into such wonderful detail. I think you're saying to just let the video software handle it and not worry about it.

I understand that real film scoring is done at 48000.

I'm assuming that adding a previously recorded cd version of a song to a video and the software then changing the sample rate must not have much of an impact on the sound quality as far as artifacts and such or it seems there would be more talk about it out there than I've found...

It was surprising to me that it wasn't all the same rate as cd quality. But I suppose people writing songs for film/tv probably already knew that.

Thanks for your help,

Josie

Hey Jim,

That is very interesting. Seems a little excessively high rates to me but that's very good to know.

I don't know if you've heard about iTunes Radio new policy and algorithm. The deal there is that if your song is mixed too hot and loud it will be automatically turned down to fit where they want the overall level of the entire play list. This move being praised by those against the loudness wars especially Bob Katz who thinks this will help get the dynamics back into music. He says under this new policy that mixing to get the loudest possible signal not only won't matter anymore but may make songs mixed too hot actually sound worse and smaller when they are automatically turned down - compared to other songs mixed with a more reasonable dynamic range to begin with. Just an fyi.

Thank you for the info,

Josie

Noel,
Most excellent graphics and definitions of the difference.

Jim,

Actually what you presented is excellent information for future.
Sheez there goes another TERA-byte HD...

Josie,

Just upgraded to Pinnacle Studio 17 Ultimate, versus vs 14 had for several years.
Am creating the music using RB with FX goodies, but 44,100 is the wav created.

However....big word, that cancels all previous words.
However, when loading into vs studio 17 editor bit of magic happens for me..

After completed editing, and touch-up, then rendering to MP4, the audio results are much
nicer sounding. They do the 41 to 48 conversion in their software. Also the new software
has lot more audio ability editing features.

End result, reasonable sounds to Video Editor....Enhanced sounds out. Am happy camper.

FWIW Disclaimer, only a user of the Pinnacle video software. It is excellent upgrade tho.
Recollections is 1920 x 1080 x 48k, with the decent sounds.

Hey Frank,

I switched to Premiere Elements 11. I also have Videostudio X6 Ultimate. Couldn't decide so I got both. LOL! Pinnacle is one of the few I didn't trial. I think most of them now have an audio editor built in. I just never noticed on my old PD10 that the sample rate was changed even though I'm sure it was....duh. So I didn't want to be doing something wrong and not knowing any better. LOL! So thank you for that info.

BTW, what kind of rendering time are you getting with Pinnacle for a 3-4 min video with effects? I know a lot depends on how many and what effects - but just an average ballpark??? My test in VS6X took around 2 hours to render to an mp4 file which I thought was rather long.

Thank you,

Josie

Josie,

I've just checked in Reaper and if I set my project sample rate to 48,000 Hz, Reaper still sees the BIAB RT files as 44,100 Hz. I assume it interprets them this way. Maybe your video software does something similar.

Try it out. You've got a great ear and that will soon tell you if there's anything wrong

For what it's worth, my song "Sunny Side of Life" has my vocals at 48,000 Hz and all the BIAB instruments are at 44,100 Hz. I rendered the wav to 44,100 Hz. I usually set everything to 44,100 Hz but this one slipped past me! Given your topic, it was serendipitous!

Here's the link to "Sunny Side of Life" if you want to have a listen.

All the best,
Noel

I don't think all this matters much because any vids going to YT gets it's audio automatically compressed down anyway. There's online articles about how to mix your audio file prior to uploading to YT to get the best results but realize those best results are far from CD quality.

Bob

Hi Josie:

I usually mix my song at 44.1 Hz which I use for WAVs, mp3s, etc. I then use Sony Sound Forge to resample the song as a 48k WAV for video work. At 48k, it sounds exactly the same as the 44.1 version, provided that I actually tell Sound Forge to re-sample at 48k and not simply SET the sample rate at 48k without re-sampling it. I believe simply setting it to 48k without re-sampling would change the pitch.

In your situation, I believe you could use Reaper to mix down both a 44.1k version and a 48k version of your song. Section 18.3, page 343 of the Reaper pdf manual talks about the different settings for rendering a project. The first window under the "Options" section is the sample rate. You could try rendering a song at both 44.1 and 48k and see if they sound any different.

I was like you -- I bought Corel Video Studio 6 AND Sony Movie Studio Platinum 12 64-bit because I wasn't sure which I could work with.
I ended up really liking Sony Movie Studio Platinum 12 because it's work flow is so much like a DAW. In fact I know several people who use the Sony Vegas (the pro version of the same software) as their DAW because it's audio features are so good.

The rendering time of 2 hours you got in Video Studio seems very long. For my first several music videos where I used still photos instead of video the render time to an mp4 file was just about as long as the song itself, maybe shorter (2-3 minutes. For the Bob Dylan parody I just did, where I used video, the render time to mp4 was about 6-7 minutes for the 2:45 minute song.

But I am using the 64-bit version of the program and I have an i-7 processor and Sony Movie Studio also lets me use my Video card (that meets certain specs) to help in the video rendering-- so I think all of that helps with rendering time.

As to why video uses 48k, I thought I read somewhere that video uses 48k for audio because you can divide 48k by the common frame rates of 24, 25(PAL) and 30 and get an even number, but can't seem find a reference for that so don't quote me on that.

Good luck.

John

iTunes isn't the only one pushing higher sampling rates. Neil Young has been pushing Pono as a platform to bring back higher fidelity to digital recordings. But for that, you've got to go back to the original tracks and remaster them at the new sampling frequency (which they are doing with Pono), or you haven't gained anything in the process.

As mentioned, upsampling means changing the sampling rate. So if you upsampled from 41kHz to 48kHz, you now have 48000 samples per second instead of 41000 samples per second. But where do those extra 7000 samples per second come from?

To a computer or media device, music is just a stream of numbers. So if you had a snippet of music sampled at some rate, it might look like this:

... .234, .300, .550, .557 ...

If this stream were upsampled to twice the sample rate, the stream would look like:

... ???, .234,???, .300, ???, .550, ???, .557, ??? ...

That is, twice as many numbers would be used to represent the same audio stream.

The computer can only guess what those values are likely to be. There are all sorts of methods which produce reasonable results (linear interpolation, cubic splines, etc.), but they're all guesses. The original information was never captured, so it can't be restored.

For example, you could linearly interpolate the missing numbers by averaging the numbers around it:

... ???, .234, .267, .300, .425, .550, .553, .557, ??? ...

So now you've got enough numbers in the stream to play it back at the new sampling rate. But you didn't add higher fidelity to the recording. To do that, you'd have to go back to the original recording and re-mix at the sampling rate.

But... If the information wasn't recorded at a higher sampling rate, the information isn't there in the recording anyway. So all your DAW can do is upsample.

That's what happens when you import tracks from BiaB at CD quality, and set your project to a higher sampling rate. Behind the scenes, the BiaB track is being upsampled.

However, all the effects and processing will be at 48kHz (possibly higher, depending on the DAW), so there is a benefit to working at a higher sample rate.

Noel,

That would've freaked me out but it doesn't seem to have had any overtly negative effects on the sound.


Josie

Bob,

Good point about YT's compression. But as well as YT, I'm making some DVD's for gifts to friends/family - hardly blockbuster material but sentimental to us. So of course I want them to sound as nice as I can in addition to whatever else I do on YT.


Josie

If I recall correctly, the most common upsampling schema is to simply fill in the "???" in DCUNY's example with zeros.

Probably the biggest 'aha' moment that I had in EE-638 at Purdue, was in the Proakis and Manolakis text when it showed how analog signals are reconstructed from digitally sampled data, through the analog impulse response superposition of the individual analog impulse responses.

This kind of sounds like gobbledegook, I understand, but just think of it this way:

A Digital to Analog converter has a certain impulse response - that is, you hit is with a little spike of voltage - the actual digitized voltage, and it rings out in a certain way in analog world. Not unlike hitting a bell with a striker.

Ever so slightly later in time, you hit that filter again with a spike of voltage, and it rings again, but it's still ringing from the previous spike.

Repeat ad infinitum.

If you add up all of the ringings from getting hit by the spikes, in other words, superposing them, you get a smooth analog response of voltage which you send to your analog amp/speakers/etc.

I've long forgotten most of what I learned in the course (1995 time frame). The long and short of it as it pertains to this discussion: the data compression algorithms with online video obliterate most of the actual real concerns of recording in 44.1kHz vs. 48 kHz, and most DAW software can handle the up/down sampling without issue or heavy hitting on the CPU.

John,

It feels good knowing I'm not the only one who couldn't decide and bought two video editors. LOL! It's like they all do different things well. Sony is a nice one too that I did trial.

I'm feeling better after all the replies in this discussion that I don't have to worry about this as much as I was concerned that I might.

Thanks for the rendering info too. My AMD computer is not that powerful for video making and has built in graphics. I know now that's not as good as having a dedicated card but it has opencl with accelerated graphics so I suppose I could've done worse. I'm gonna do a test video on PE running on 64bit and see if it makes a difference in rendering time compared to VS6X at 32bits.

Thank you,

Josie

Hey David,

That was a great explanation. I'm going to read it again and get back on here later.


You too Scott!

I'm tired now so I'm gonna process what you and David are saying a bit more.


Josie

Rockstar_not is correct regarding the upsampling process. To also state the obvious, if you record at 48K, the recording has to be dithered down to 44.1k to print a redbook CD.

In the most recent edition of Bob Katz book on mastering and recording, he recommends recording at 48k 24bit.

Simple.

I record at 24 bits and 44.1k sample rate.

I render waves to 16/44.1k and let it roll.

The main reason being it's the industry standard.

I do understand all the arguments for and against higher sample rates. I just don't have the time to worry or ponder those things in depth. I listen to the song at 16/44.1 and like what I hear.

That's good enough for me.

The only time I concern myself with sample rates is with MP3 conversion and then I opt for the highest rate of 320kbs and once again, let it roll

I record and render with the identical specs as Herb does for the same reasons. Most all of our music is listened too via CDs or MP3s.

FWIW I think recording and/or rendering anything over 24/48 is just a waste of disc space for the genres of music that we are recording. If one were doing an orchestral piece then maybe higher bit and sample rates may pay off if you have golden ears. I do not!

Remember if it sounds good it is good!

Hey all, I'm back.

Josie. Sampling rate comes down to the Nyquist Theorem. Basically what it says is that the highest frequency that you can record is half the value of the sample rate. So, with 44.1K, the highest frequency you can record is 22.050KhZ. Human hearing goes to about 20kHz. MINE goes to about 14 or 15kHz, on a good day! My cat's may go to 60kHz. For 48k, your upper sampling frequency will be 24kHz. Out of the hearing range of humans, but still able to capture harmonics of a sound. Then we get into 88.2, and 96kHz sampling rates, which are 'double speed.' My cat could possibly hear sounds that high, but I can't. Most microphones and speakers will not reproduce sounds that high.

Then we have the idiots who want to record at 192kHz, or now 384kHz. WHY??? There is no audible sonic material at those levels, plus any filtering is going to drive the noise floor UP, not down.

I record at 24bit, 48kHz. That gives me plenty of dynamic range, and a sampling rate that doesn't give me overly huge files, but still captures all the details I want.

I'll add more to this tonight when I get home.

Gary

There was a guy over at FAWM this year that had wonderful sounding recordings and they were super loud! But they didn't seem overly compressed at all. I asked how he got things so loud, but so clean and he said:



Recording at the higher resolutions is all about having "headroom" to compress without affecting dynamics. Sort of off the topic of video vs. audio, but I always feel the need to interject something -- especially when I don't know what I am talking about (ha, ha). I switched to 44.1/24 bit recording and I haven't noticed any differences yet (LOL!).

EDIT: Here's a link to his fawm page: http://fawm.org/fawmers/lenigmusx1/ and here is one song: http://fawm.org/songs/38913/

Josie,
To continue, when you record, you have two things you need to worry about. Sample Rate, and Bit Depth. Today, most recordings are done in either 16 bit or 24 bit, although you will see 32 bit floating once in a while.

Think of a graph for me, please. As we move from left to right in the horizontal plane, we increase our sampling rate. Along with the sampling rate, our highest frequency that can be recorded increases as well. So, as I said before, Nyquist says that the sample rate must be twice that of the highest frequency sampled. Sampling at 48kHz gives you a 24kHz frequency range.

If we move up the graph, we increase our bit depth, which equates to our dynamic range. Remember, Dynamic Range is the range, in decibels, from dead quite to the loudest sound we can record in digital. Each single bit gives us 6dB or range. So, an old 8 bit recorded would give us a total of 48dB of dynamic range. My conversations are in the 55-70dB range, depending on how agitated or excited I am. A standard CD has a bit depth of 16 bits, giving you 96dB of Dynamic range. Twenty four bits will give you a dynamic range of about 144dB, but the thing is, you get into electron flow noise at that level. The gear I work on, our top of the line A/D converter, will top out in the -127 to -130dB range. That's a pretty substantial noise floor, and that's without A-Weighting.

Let's go back and discuss these numbers some, okay?

A rock concert may have SPL, or Sound Pressure Levels in excess of 140dB! That is actually damaging to hearing, but they still do it. Now, if you're recording at 16 bit, you have a 96dB dynamic range. How do you compress 140dB into 96dB? Well, that's pretty easy, actually. You attenuate the input level so that at the maximum SPL you're getting to the maximum of the converter, which is this case is 96dB. Ahhh, but what about the really silent parts? In a straight recording, they may be lost. Subtract 96 from 140, and you have 44dB. Now, that 44dB is your noise floor, not zero. Anything below 44dB is lost. This is where compressors come into play. If you have a compressor before the inputs to your A/D converter, you can squish the sound to get all 140dB of the concert range into a 96dB dynamic range for a CD.

Okay, none of this answered your question earlier, though. If you upsample a 44.1kHz, 16 bit signal to 48kHz, not much is going to happen. You're adding frequency range and bit depth, which are basically zeros. Downsampling, for most people, you will not hear a difference. There are some who will, but most, it's doubtful. Remember that most recordings today are recorded at 96/24, and then downsampled to 44.1/16 for CDs.

I hope this helps you a bit.

Gary

Charlie, Herb and Mario, thank you guys, I appreciate all of your insights.

Kevin, Thanks for the info and the link. I really enjoyed listening to that guy.

Gary, Welcome back! Hope you are doing well. We've missed you around here. Thanks for the info.

I think I have a much clearer understanding thanks to all of you. You guys really are great.

Josie

Hi Josie. I stumbled on this thread while I was searching to 'sample' a 'bit' (pun intended) of your music postings. The point being I stumbled onto "She's in Love" and absolutely loved it. Around here, its known as beach Music and remains popular even today because it creates the beat and sound of the SC state dance, The Shag. There are two local Myrtle Beach stations that feature classic and beach music and play and feature local artists recordings and I'm sure if they had this tune, they would be all over it. You can find streaming of one, 105.3 The Surf on line. You did a really good job on writing, arranging and producing "She's in Love". It's one of my favorites of all the posts I've listened to since joining the forum.

Regards,

Charlie

Josie, thanks for the welcome back. I'm alive, I'll give you that. Things have been better, but hey, I just upgraded to BIAB 2014, so I've got a lot of catching up to do. I would appear, young lady, that you are doing VERY well. I'm so happy for you. Keep it up!

Gary