Syntrillium Backup Forums :: View topic - File format and compression for radio download

aglady · Posts: 8

I am a new CEP user and produce news interviews for a government agency in my state. I was able to upgrade from analog, but because of budget woes had to teach myself digital audio with no training. So I have come a long way, baby. But have a lot further to go.

My question is not -- Can you take out the vocals -- although I've already got that too -- but how to reduce my file sizes for faster downloads by my partner radio stations?

I currently record a 3 minute voice interview in .wav, compress to .mp3 and then zip with win.zip before uploading. Because of costs, I am very limited as to the server space my agency currently allows me. Also, some of my radio station partners are in the boonies and have only a connection that is as efficient as two cans strung with string. A three minute cut may take 1 hour to download.

My biggest question is what sampling and bit rates are necessary for there to be high enough quality for radio? Is there a minimum sampling rate? Is there a difference for stereo v. mono? Is there a difference for FM v. AM? What do stations require? Is there such a thing as too much quality – when file size is an issue?

Thanks in advance. I have learned a lot reading the BBs.

Wildduck · Posts: 466

OK, I'll jump in. Having read this through and realised how confusing it looks, make sure that you understand the difference between sampling rate of a file and the compressed bit rate.

It very much depends on what the receiving site can cope with and how they will play your contributions.

With .mp3 coding, 16 bit mono files with a sampling rate of 32k compress quite satisfactorily down to 64kbits/sec or even down to 56k. These are speech files of news interviews.

At these bit rates they can be sent over direct ISDN or even modem links in more or less real time. A sampling rate of 32k samples/sec gives an audio bandwidth of 16kHz. FM transmitters normally filter out frequencies above this, so audio at this sampling rate ought to be 'good enough'. This gives a 10:1 compression ratio. A 1 minute uncompressed mono file is approx 4 megs in length at this sampling rate. After compression it becomes 400kBytes.

Stereo files, being twice the size, need to be at least 128kbits/sec when compressed.

If the stations want you to use .mp2 encoding (and you can locate the appropriate plug-ins), it is usually said that a higher data rate is needed.

If you and the stations have broadband connections, straight internet file attachment transfer or posting them on your website might be the way to go.

Always remember that the person sending the file is sending off their 'baby'. But to the person who has to receive it, it is probably just a load more hassle and work. Make it easy at the receiving end.

Hope this helps.

aglady · Posts: 8

:)Thanks, Wildduck. I can see that I could use an engineering degree. I wish I could get my IT folks to see the need for putting our stuff out in as user friendly form as possible. Well, maybe after our state lays off another 50,000 workers as predicted -- there will be more $$ to work with for the rest of us, or more likely there won't be any of the rest of us needing to ask such questions!
Thanks
AgLady

Craig Jackman · Location: Canada Posts: 909

To condense what Wildduck had to say, I'd recommend the following to get the smallest possible file size yet still be listenable:

- use mono files so that your file is half the size of a stereo file

- record at a sampling rate of 22kHz rather than CD standard 44.1. If you are just doing voice recording, this will be on the bottom edge of acceptable, but it will give you a smaller file size.

- I can't recommend recording in 8-bit as that sounds brutal, but do record in 16-bit as opposed to 20, 14, or 32-bit.

- If you already in mono/22kHz/16 bit, then use as small a bit rate as you can in MP3 conversion that will still leave you a listenable file. Personally, I wouldn't go below 96kbs, but if you want the smallest file go to 64 or 56kbs.

It is important to remember that the important end of this equasion is the receiving end. If your end users are ****ing and complaining that it's taking them an hour to get your files (they really should upgrade their modems and ISP to get the fastest service possible, but I digress ...), then phone it in. Then they are recording in real time. Can't phone x-number of stations at once? Then investigate getting a service that will stream your reports, like Real Player. Maybe setup a dedicated phone line that plays the reports off an answering machine. Sure the recieving stations want is to sound better than phone, but if they can't or won't make it easy on themselves, you have to find another way to make it easy.

aglady · Posts: 8

:kiss:Thanks, Craig. That is just the info I needed. Saving even 1MB per file can mean a lot when you are talking about 20 files on the server at a time. And I take your comments about providing easy to use cuts in a hassle-free format. It helps us all to do what is best for the group -- Hey, that's the the brillant new theory from "A Beautiful Mind". You should get the nobel!
Aglady

Chris Diack · Location: New Zealand Posts: 2

Just a thought... As you're dealing with many stations and you're wishing to provide them with the most satisfactory product at the most cost effective price, keeping yourself within budget, and keep yourself employed, the following suggestions may be of help, in the light that you're relatively new to the business... Top and tail the audio cuts so you have no dead air at the start and the end of the audio... Normalize your audio to the same standard every time; We use 85% (-1.4 db) as a standard but in order to be able to import the file straight into an automated process it may be desirable to make that level even less... How we deal with this situation in our production unit is to simply ask the client to send a demo file back to us and we adopt the same standard... We also apply a little limiting, generally around 3db to tidy up the levels a little, as unless you're doing this sort of thing all the time, you may find that the levels are either low at the end or low at the start of the recording depending on your confidence level and therefore projection of your voice... These are just small things but may help you consolidate your position and job which is no doubt very important to you... If you need any more guidance, I'd be delighted to assist, if you wish to send a copy of one of your files... I am a semi retired news editor and presenter of 30 odd years and enjoy coaching people who want to improve themselves... Regards, Chris Diack, New Zealand

the original radiogeek · Location: USA Posts: 8

I've found a better way is not to send the file, but to send a link. If you've got the capability, post the file to website or an FTP site, and send the link to the recepient. I use this technique with our sales people. I used to send MP3's of the final commercial mixdowns to our sales folks, but it quickly clogged their e-mail accounts. Plus, some accounts have very limited space, and will not take files over a certain size. I asked our tech guy to set me up with an FTP site. I now post the MP3 to the FTP site, and e-mail the link. Takes much less space, and they still get the file. The only down side has been that they can only open the file, not right-click and "Save as". I'm guessing that's just because of how our site is set up. Don't know if that helps, but I thought I'd throw it out there.

aglady · Posts: 8

:)Thanks Chris and Radiogeek. I will try some of your suggestions. I have not done much with normalizing other than test it, but have seen other produced pieces that obviously use it.

Chris: I already put my cuts on our website and email the link to our users. It allows them to download only what and when they want to use our material. Because we upload a zipped .mp3 our users have the option to "save" or "open" the file when they chose to download. It gives them more flexibility, but it does mean another step if to unzip to .mp3 and then convert to .wav.

Appreciate the help. You guys are great.

aglady · Posts: 8

Craig: I have taken your advice and have been playing with the bit and sampling rates. I have tried mono/22mhz/16bit recording, but am unable to convert to a .MP3 of 96Kbps. CEP only allows a top rate of 56Kbps when converting from a 22mhz file. Also this is a CBR (constant bit rate) conversion in a standard .MP3 encoding. I am not even able to select the .MP3Pro encoding option.

I can convert to a 96 Kbps .MP3 if I record in 32000Mhz but again only in the standard .MP3 encoding. The .MP3Pro encoding option only alows a conversion to a top rate of 48 Kbps.

Would you suggest recording at 32000 or 44100 so that I can compress to 96Kbps? It seems that .mp3 compression has the most to do with file size. Since they are all factors in size, where should I make the trade off?

As you can see, I am at the edge of my understanding of all the engineering specs., but it seems, in terms of final file size, the sampling rate has little to do with it. It is the .MP3 conversion rate (eg. 56 vs. 96 Mhz) that cuts my my file size in half. My ears really don't register a difference -- and in audio, that is what I've always trusted.

Any last wisdom on the subject? I have read many of you postings and I want to praise you on your thoughtful, respectful Smile

comments. I fail to understand some of the nasty:( replys some users give. We are all just trying to learn and improve. Thanks for being supportive, even to the CEP challenged!

Aglady

Craig Jackman · Location: Canada Posts: 909

If file size is the primary concern, then record with a sampling rate of 32k. Since these are news features, and mostly voice, I doubt that most people would notice any difference in sound quality.

You are doing absolutely the BEST thing, but trying and trusting what you hear. If you're OK with 22k/56kbs files, then do that! You certainly will have small files. Just make sure that your recieving stations are capable on those specs. There's not a lot gained by the small file size if they have to spend the time they just saved converting the downloaded file into something they can use.

There are 4 elements in files size. Bit depth (8, 16, 20, 24)- but you don't want to use 8 bit as it sounds bad. 16 bit is standard, higher bit numbers are for audio geeks. The smaller the bit depth the smaller the file. Sampling rate (22k, 32k, 44.1k, 48k and mulitples of those). Again, smaller the number, the smaller the file size. You won't see a big difference in the file size between 48 and 44.1 or 44.1 and 32, but you will see a difference between 22 and 44.1. 44.1 is the CD standard, 32 is about FM radio quality, 22 about AM radio quality. Again, big numbers are for audio geeks. Mono vs. Stereo Mono files are half a stereo file, so it will be half the file size. Lastly, MP3 bit rate. MP3 works by throwing away what it thinks you can't hear. The bigger the number, the more of the original file is kept. Even at the highest bit rate, 75% of your original file is gone. What you have to decide is what balance you want between file size and sound quality from your original file. That decision would be best made in consultation with your recieving stations, as you can make a teeny weeny file that's virtually unlistenable, or a great big file that sounds great.

aglady · Posts: 8

:kiss:You da man! I know it is a bit tedious to deal with us non-radio geeks, but you have really helped me begin to understand something that I don't think the general audio "joe" or "joanne" understands.

I am working with my receivers, but they don't always know either. So again, thank you for your supportive comments. You are very knowlegeable and are generous in your help. We (all the forum users) should be as kind and admirable.

Aglady is in your debt!

Wildduck · Posts: 466

I'd say stick to mono, 32k sampling rate and the approx 10:1 compression ratio that produces the 64 or 56kbps mp3 files.

Then try to find someone at at least one of the receiving stations that understands what you are talking about and try to sort out the simplest way that they can receive the files. If you get this far, then try to talk to the other receivers about what they need from you to make things easy. I know I said this before, but I can't stress too much how you must aim at making life easy for the receiving stations.

Making any inter station system work does really need someone who can suggest sensible approaches and can appreciate the possible problems. That's why I am in a roundabout way suggesting that you try to befriend someone with some technical background somewhere on a station to help with advice. The technology is not too difficult. The overview of the 'system' is where the work has to be concentrated.

If you want to discuss specifics off-list, I've had a lot of experience with sorting out just this type of arrangement.

aglady · Posts: 8

Thanks, wildduck. I am doing just what you have suggested and am mirroring what one of my station partners produce for their network. I have noted that CEP doesn't allow a 22mhz mono file to convert to a 96Kbps .MP3 as suggested by Craig above. It does allow that conversion with a 32,000mhz mono file. I am finding that my partner station uses 128Kbps, which really eats up my file size limits, but the quality is definitely better. I feel like I am getting to where I need to be and a 2MB file size is livable. Station feedback has been very positive, so must be on the right track. Thanks:)
Aglady