Converting Audio for Phone Hold Music

josh88

Remarkably Tired.
Fight Leukemia
I've been asked to convert a bunch of music for our season over to be used for our theatre hold music. Our phone provider wants them as wav files. No problem. They also want the files to be:
-PCM16 format
-16.000kHz
-16 Bit-Mono Sampling


Also no problem. BUT they also want the files to be no larger than 4mb. Thats where I'm having the problem I can convert this stuff all day long but for a full length song I can't do anything to get the files as small as they want them. I've run through all my usual programs and I've tried all kinds of online converters too just to see if it makes any difference and I just can't make them small enough without compromising somewhere, usually lowering the bit rate. That makes it sound terrible.... but also sounds like pretty much every other hold music I've ever heard. Is that just what I have to do or does someone have an idea I'm not thinking of?

They're reaching out to our provider for more details but I haven't heard back yet. I"m assuming the 4mb limit is the hard line because they can't use anything larger, I have a feeling the other requests are for their "optimal" file but that bit rate and the sampling size aren't hard limits.
 
Split the file up til it’s only 4mb.

There might be a pause between files but such is life. You could tell them to throw in taking bits in between.

Personally nobody should be on hold for an entire song anyways. Bad CSR imo.
 
Since the audio files you're describing are not compressed in any way, the file length is directly proportional to the playback time. A 4 MB file at 16 kHz, 16 bit mono sampling works out to 125 seconds. You just can't fit more than a couple minutes into a file meeting those constraints, any more than you can drive more than 130 miles in two hours without exceeding a 65 mph speed limit.

Tinkering with the bit rate is (by definition) going to change the file format to be something other than what they want.
 
Split the file up til it’s only 4mb.
There might be a pause between files but such is life. You could tell them to throw in taking bits in between.
Personally nobody should be on hold for an entire song anyways. Bad CSR imo.

I agree nobody is ever going to hear a whole song and I may just do that because nobody handling the files other than me will know they are split up.

Realistically I know there is no way its possible with the files they have given me and the constraints. They're just insistent that "well we did it last year, there must be some way" Just doing my do diligence and making sure I wasn't missing some magic process that would do what I need. While keeping them uncompressed and any kind of quality.
 
Nope. If there is though let me know lol I want in on the ground floor of that discovery.
 
I agree nobody is ever going to hear a whole song and I may just do that because nobody handling the files other than me will know they are split up.

Realistically I know there is no way its possible with the files they have given me and the constraints. They're just insistent that "well we did it last year, there must be some way" Just doing my do diligence and making sure I wasn't missing some magic process that would do what I need. While keeping them uncompressed and any kind of quality.

Oh, this is about size? Because NOTHING regarding music on hold is about quality. Start with a band pass filter, 300Hz on the low and about 3kHz on the top. That's the *equalized* bandwidth of a POTS connection. Now when you do that edit, also introduce some wow & flutter effect to mimic an old tape loop. Telescope (edit out repeats, long music bridges) down to 2 minutes and change and you're set to inflict this on an unsuspecting world.

Does the Valentine's Equity contract and copyright clearance licensing allow you to use recordings in this way?
 
Does the Valentine's Equity contract and copyright clearance licensing allow you to use recordings in this way?
Yes, the music we're using has been cleared, and we received the files from the producing companies of the various tours and shows for this purpose.
 
Yes, the music we're using has been cleared, and we received the files from the producing companies of the various tours and shows for this purpose.
:D:D:D
 
See if the phone system will accept an 8 kHz sampling rate. It'll still sound fine and the time per file will double. Otherwise, do one song per file, fade out and end each one at 2 minutes.

Some phone systems will do MP3 files with an optional license applied. Of course, they charge for the license.
 
This is interesting. I used to install small PBX systems as a gig in high school. My boss had a stock of radio shack AM/FM radios and 1/8" TRS to RCA cables to connect to the hold music in to the phone switch. I ziptied many a radio to a backplane in my day. I guess that's obsolete now?
 
This is interesting. I used to install small PBX systems as a gig in high school. My boss had a stock of radio shack AM/FM radios and 1/8" TRS to RCA cables to connect to the hold music in to the phone switch. I ziptied many a radio to a backplane in my day. I guess that's obsolete now?
I'm not even sure we have any option like that, as far as I know its all being handled on the phone company side. We're giving them all the files and they're uploading them for however our service works. It would be much easier if our hold was actually just some kind of transfer to a playback device with whatever content playing on a loop but thats all above my pay grade and not something I want to make myself the volunteer manager of.
 
Oh, this is about size? Because NOTHING regarding music on hold is about quality. Start with a band pass filter, 300Hz on the low and about 3kHz on the top. That's the *equalized* bandwidth of a POTS connection. Now when you do that edit, also introduce some wow & flutter effect to mimic an old tape loop. Telescope (edit out repeats, long music bridges) down to 2 minutes and change and you're set to inflict this on an unsuspecting world.

Does the Valentine's Equity contract and copyright clearance licensing allow you to use recordings in this way?

Respectfully disagree with the procedure mentioned here, but that's only because I worked in industries where I handled low res audio for answering machines and public announcement systems and that kind of thing all day long for many years.

Do this.

1). Start with your high res audio file - 44kHz, 16 bit stereo. Convert to mono but be aware of the issues with stereo to mono MUSIC conversion - sometimes the mix balance changes (modern tracks are not mixed for mono compatibility).

2). Agree with cutting some lows. Not necessarily 300Hz, maybe a bit lower - like 100-200Hz.

3). Do not cut any high end at this point. Do not apply any low pass filter.

4). Make sure your 44k-16 audio is compressed and limited to the point where you are making the best use of the dynamic range. Many limited playback systems, especially those that implement audio data compression do not do well with audio files that are low in signal level. That tends to add significant noise to the end product. You want to avoid this. With pop music not so much of a problem but with broadway stuff and orchestral stuff there is too large of a dynamic range. Need to compress that. Do not worry about ruining the music by compressing it (as you can do it to taste and no need to consider this audio will be pristine when done). It is more important that you hit the sample rate conversion with enough level because it's going to cut quite a bit of high frequencies by nature of the anti aliasing filter and hitting that level harder means you lose less high end, allowing you to potentially recover more. If you want to minimize the effect on a mixed music track then use a buss compressor, like a Fairchild 670 emulation or Waves Renaissance Compressor with opto style buss compression. Be sure to limit your track after, with something like Waves L1. Conversely, make sure your file is not too crispy - if there's 3dB to 6dB of visible headroom in your file between peaks and RMS that's probably the butter zone.

5). Sample rate convert using editing software - Adobe Audition, Izotope, SoX, etc. Use the best converter you can get your hands on. Make sure anti alias filter is engaged.

6). Now you've lost some high end. AFTER conversion to 16kHz open your wave file in an editing application that will be able to save the file back out at the same sample rate without upsampling/downsampling (Bias Peak, Audition, etc,) NOT Pro Tools, Nuendo, Cakewalk, etc.. Give the audio a slight high frequency boost, like +2dB shelf at 3kHz. Yes, at 16kHz. This will bring back some of the high end you lost in translation. Re-apply limiting if need be.

7). Now, make musical cuts and edits in your track to get it down to your target size. If you have to cut a chorus or a verse, no prob. More important that it fits your memory footprint than the piece of music be exactly as it was when it started. If you had a playback platform capable of repeating audio files you could cut this into chunks where you can repeat a chorus, etc.

Your end playback medium (telephone) may not support frequencies higher than 4kHz in theory, but by doing this you are providing some pre-emphasis boost and you will make the most of the frequency bandwidth and dynamic range you have available to you in your end platform. Don't forget many end user platforms have low pass filtering already built in, you can attempt to guess and pre-filter the high end out but at the end you may discover you can get more HF out of the end system without cutting the high end out of your files.

I hope this helps. This process has worked for me for many years. You should be able to attain much better results using this method.
 
Last edited:
Respectfully disagree with the procedure mentioned here, but that's only because I worked in industries where I handled low res audio for answering machines and public announcement systems and that kind of thing all day long for many years.

Do this.

1). Start with your high res audio file - 44kHz, 16 bit stereo. Convert to mono but be aware of the issues with stereo to mono MUSIC conversion - sometimes the mix balance changes (modern tracks are not mixed for mono compatibility).

2). Agree with cutting some lows. Not necessarily 300Hz, maybe a bit lower - like 100-200Hz.

3). Do not cut any high end at this point. Do not apply any low pass filter.

4). Make sure your 44k-16 audio is compressed and limited to the point where you are making the best use of the dynamic range. Many limited playback systems, especially those that implement audio data compression do not do well with audio files that are low in signal level. That tends to add significant noise to the end product. You want to avoid this. With pop music not so much of a problem but with broadway stuff and orchestral stuff there is too large of a dynamic range. Need to compress that. Do not worry about ruining the music by compressing it (as you can do it to taste and no need to consider this audio will be pristine when done). It is more important that you hit the sample rate conversion with enough level because it's going to cut quite a bit of high frequencies by nature of the anti aliasing filter and hitting that level harder means you lose less high end, allowing you to potentially recover more. If you want to minimize the effect on a mixed music track then use a buss compressor, like a Fairchild 670 emulation or Waves Renaissance Compressor with opto style buss compression. Be sure to limit your track after, with something like Waves L1. Conversely, make sure your file is not too crispy - if there's 3dB to 6dB of visible headroom in your file between peaks and RMS that's probably the butter zone.

5). Sample rate convert using editing software - Adobe Audition, Izotope, SoX, etc. Use the best converter you can get your hands on. Make sure anti alias filter is engaged.

6). Now you've lost some high end. AFTER conversion to 16kHz open your wave file in an editing application that will be able to save the file back out at the same sample rate without upsampling/downsampling (Bias Peak, Audition, etc,) NOT Pro Tools, Nuendo, Cakewalk, etc.. Give the audio a slight high frequency boost, like +2dB shelf at 3kHz. Yes, at 16kHz. This will bring back some of the high end you lost in translation. Re-apply limiting if need be.

7). Now, make musical cuts and edits in your track to get it down to your target size. If you have to cut a chorus or a verse, no prob. More important that it fits your memory footprint than the piece of music be exactly as it was when it started. If you had a playback platform capable of repeating audio files you could cut this into chunks where you can repeat a chorus, etc.

Your end playback medium (telephone) may not support frequencies higher than 4kHz in theory, but by doing this you are providing some pre-emphasis boost and you will make the most of the frequency bandwidth and dynamic range you have available to you in your end platform. Don't forget many end user platforms have low pass filtering already built in, you can attempt to guess and pre-filter the high end out but at the end you may discover you can get more HF out of the end system without cutting the high end out of your files.

I hope this helps. This process has worked for me for many years. You should be able to attain much better results using this method.

I guess the inclusion of Wow and Flutter was not a sufficient pointer to the tongue in cheek.....

After years of listening to badly done tape loops, radio stations injected at distortion levels, and "in house" stuff that made me want to hang up rather than wait for my party to answer.... I was hoping to duplicate the "experience" for those who have spent their lives counting ones and zeros.
 
I guess the inclusion of Wow and Flutter was not a sufficient pointer to the tongue in cheek.....

After years of listening to badly done tape loops, radio stations injected at distortion levels, and "in house" stuff that made me want to hang up rather than wait for my party to answer.... I was hoping to duplicate the "experience" for those who have spent their lives counting ones and zeros.

:) YES!
 

Users who are viewing this thread

Back