I uploaded a recording of a phone call that I taped to a web page(me calling the Subway helpline trying to make reservations :P). I used a program called Sound Forge to record them though my sound card’s line-in. There were all these damn formats - mp3, wav, wma, and rm were the one’s I recognized. Of course I first saved in .rm format not realizing that noone fricking used it. I uploaded this one first and it was a very small file. Since Sound Forge could not load an rm file for editing (how dumb is that?) I redid the whole thing in mp3 format so I would have a more viable format. The file was like 8 times the size even at 22 odd mhz, mono. When I played these files off the web page, it stuttered, I suppose because there was so much more data. The quality however wasn’t any better (when it wasn’t stuttering). None of the formats seemed to be as small as rm. I would use rm if it was used by any of these editing programs but it really isn’t. I also want it to be able to be opened by many types of players. What format should I be using and what setting (mhz, bit rate) would not be overkill for the quality of a recorded phone call?
The size of the file is determined by the bitrate, measured in kbps. The samplerate, measured in khz, determines what sound frequencies can be heard in the file. I would suggest you use at most 22Khz, possibly even 16Khz.
My first choice would be to use MP3. It has the best quality generally, and is by far the most compatible. Here is a link to the WinLAME MP3 Encoder. It will produce files of much higher quality than the MP3 encoder that comes with Soundforge, and should be rather simple to use. Just install it, run it, choose the files to encode, set your options up, and click the Finish button. I would suggest you set the Quality to “0, Slowest and best”, set the frequency to 16000Hz, and set the bitrate to an Average of 32kbps. Set the number of channels to Mono, since it’s just a phone conversation. If you want better quality, try Average 64kbps, and set the frequency to 22000Hz.
My second choice is WMA. It generally has better quality than RealMedia, and doesn’t require most users to download any player application. 32kbps/22Khz/Mono should be fine, or you can try 64kbps/22Khz/Mono if you want better quality.
If you’re aiming to let dialup users download the files, you’ll want to set the bitrate to at most 32kbps, no matter what format you choose, or maybe even as low as 24kbps. If you can’t figure something out, feel free to reply and I, or someone else, will probably answer your questions in short order.
MP3 is a streaming format which carries unnecessary overhead information for a contained file. You can get exactly the same information and quality encoded with the same MPEG Layer3 (MP3) codec in a WAV file which will be somewhat smaller. For voice you can save a WAV at 11025, mono, 20 KBits/S, and it should be of reasonable quality. For music you’d obviously want higher quality.
IANAL, but isn’t against the law in some states to record people without telling them? You might not want to post this call on the internet.
You sure about that? Sounds to me like you’re describing a WMA file, as I’ve never run into a wav file that is smaller than a corresponding mp3.
Personally, I don’t like WMA because it supports DRM. I don’t like Media Player for the same reason. I’d suggest the mp3 encoding, which can be done at similar settings as were mentioned by sailor.
First off I have to repeat what Rhum Runner said. Some states have laws about recording someone without their okay so you might want to check on that.
About sound file formats: mp3 is good because 99% of computers and OS’s can play mp3s. Same goes for wav files but they tend to be bigger. I dislike wma and rm files simply because they need special programs to play them. Macs are out of luck with wma and Unix people have to jump through hoops to get them to run (and I just don’t like Real).
sailor: Whether it is a streaming format or not doesn’t make any difference as far as quality. ALL audio formats you encounter today are streaming formats (RealAudio, Vorbis, MP3, WMA, Quicktime). Embedding an MP3 into a WAV will INCREASE the file size by an amount equal to the size of the WAV header and offer no benefits. The MP3 audio is simply being packed into a different container, unchanged.
asterion, yes I am sure. It has been discussed in old threads several times.
FDISK,
>> Whether it is a streaming format or not doesn’t make any difference as far as quality
Whish is exactly what I said.
>> ALL audio formats you encounter today are streaming formats
NOT true. You’ve never encountered a WAV file?
>> Embedding an MP3 into a WAV will INCREASE the file size by an amount equal to the size of the WAV header and offer no benefits. The MP3 audio is simply being packed into a different container, unchanged.
WRONG! This has been discussed before so I will not go into details. You can do a search. Even more: why don’t you actually try it before saying something so wrong?
A WAV file encoded using the Layer 3 codec contains EXACTLY the same useful information as the same fileencoded to MP3. Exactly the same, so the decoded output is exactly the same.
The difference between the two is that the WAV file records the encoding information only once in the header while MP3 repeats the encoding information along the length of the file thus making it larger. This allows you to start decoding at any point even if you do not have the header.
Please see earlier threads where this has been discussed at length. This topic comes up periodically.
I will add WAV files are easier to process as there are more programs out there which can handle WAV than MP3. You can encode a file to MP3-L3 with SNDREC32 which comes with windows.
This site has a good state-by-state breakdown of laws about taping conversations.
Great site Number. I hope KidCharlemagne takes the time to look.
sailor: Allow me to rephrase, all COMPRESSED audio formats one encounters are streaming formats. One almost never sees WAV files online or in programs, simply because they’re too large to be practical.
And no, embedding an MP3 inside a WAV header DOES NOT decrease the size. The MP3 file itself is maintained UNMODIFIED, it is simply wrapped in a WAV header. As such, the size of the file will increase by an amount equal in bytes to the size of the WAV header. If you were to remove the header and streaming information from an MP3, it would NO LONGER BE AN MP3. It wouldn’t be an ISO MPEG1 Level-3 Audio stream, and it would no longer play in a “real” MP3 player. Stripping off all the MP3 header information, invalidating the file, and cramming it into a WAV header just so you can shave a few hundred bytes off the file while being required to use a proprietary player is just silly.
And yes, I have done wrapped MP3s in WAV headers many times myself. It’s how you get an MP3 inside a video container such as AVI using less than intelligent video editing software. Every time, the resulting file is larger by a factor equal to the size of the WAV header, or several KB. There, I just did it again. File size went up by exactly 170 bytes.
FDISK, You are quite mistaken and, as I said, this has been discussed several times in the past. Just look it up.
>> Allow me to rephrase, all COMPRESSED audio formats one encounters are streaming formats.
Speak for yourself. I find plenty of WAVs online. In any case, this is irrelevant to the compression issue we are discussing.
>> One almost never sees WAV files online or in programs, simply because they’re too large to be practical.
You are, obviously quite confused when you imply a WAV file cannot be compressed. A WAV file can be raw PCM or it can be compressed with any number of codecs, the MP3 codec being one of them.
>> And no, embedding an MP3 inside a WAV header DOES NOT decrease the size. The MP3 file itself is maintained UNMODIFIED, it is simply wrapped in a WAV header.
You are very wrong and you have no clue. A WAV file can be compressed using the same codec. It is NOT an .MP3 file “wrapped” as WAV. I wish people would not argue things which can so easily be disproven by anyone with access to a computer: Open an uncompressed WAV file using SNREC32.
- File/save as/ Format Change (you get a new window titled sound selection)
- Format: MPEG (select from the pull down list)
- Attributes: (select bitrate etc from the pull down list)
- Save file in new compressed format.
Any audio processing program, including CoolEdit which I use, will allow you to save WAVs compressed with any number of codecs you may have installed in your computer including MP3, DivX, etc.
>> As such, the size of the file will increase by an amount equal in bytes to the size of the WAV header. If you were to remove the header and streaming information from an MP3, it would NO LONGER BE AN MP3. It wouldn’t be an ISO MPEG1 Level-3 Audio stream, and it would no longer play in a “real” MP3 player. Stripping off all the MP3 header information, invalidating the file, and cramming it into a WAV header just so you can shave a few hundred bytes off the file while being required to use a proprietary player is just silly.
What is silly is talking about what you do not understand. You do not need any special player to play an MP3 compressed WAV file. It will play on all windows media players, SNDREC 32 and pretty much ANY player. Even more so than .MP3. Can you tell me of any player which will not play MP3 compressed MP3s if you have the codec installed?
I would appreciatte it if you would inform yourself by reading past threads and doing the experiment I just said. I get really tired of explaining the same things over and over and over to people who just will not listen and just will not verify the obvious.
My other pet peeve is having to explain over and over that Systray is not in any way necessary to run Windows. The other misunderstanding that will not die.
I have placed an MP3-compressed WAV file here so you can check it out. The original, uncompressed, file is 60 seconds @ 11025 = 661500 samples @ 16 bits = 1323000 bytes. After compression at 16 Kb/sec the size is down to 119566 bytes (less than 1/10th).
Being WAV the compression information is stored only once in the header. If it was .MP3 the compression information would be repeated periodically along the file which would make the file longer but the effective audio information would be exactly the same (assuming same bitrate etc) because it was compressed using the same codec.
Practically everybody has the Fraunhoffer MP3 codec as it comes with Internet Explorer and other software. It is the same MPEG-L3 codec for .MP3, WAV or any other format which cares to use it.
I have also used the DivX audio codec as I think it gives better compression but many people do not have this codec.
And to answer the OP: I have not used Sound Forge but I bet you can save in MP3 compressed WAV and in many other formats. If not you can use SNDREC32 which comes with windows. The trick is to know what parameters to select. For plain voice you can start off with uncompressed 11Khz, mono, 16 bit and select a compressed bit rate between 16 and 24 Kb/s. The file I linked to is 16 and sounds reasonably well. You can click on properties of that file to see the parameters. In summary, the trick is to choose the correct parameters and you can get the feel by experimenting. Rather than choosing each time, I have three standard settings I always use, low for voice, medium, for low quality music and high for hight quality music. There is no point in saving in high quality a piece that was originally recorded in 1927 in mono and with atrocius quality.
And I will add that RealAudio is a format which many people, among them yours truly, refuse to use. WAV (compressed or not) has the highest compatibility.
sailor: Again, you’re simply wrapping an MP3 file in a WAV RIFF header. Rename one of your MP3-containing WAVs to MP3. What happens? It will open and play as an MP3, and WinAMP will even let you add and edit ID3 tags, AND the file will stream properly. Feel free to test this yourself, I did.
I just recreated your experiment using a decoded version of your source file. I used the LAME ACM codec and SNDREC32, and the file came out 300bytes larger than yours. Why? This is simple variation between MP3 codecs. LAME compresses slightly differently than Fraunhoffer. Then, I used the straight command line version of LAME with the --mp3input tag, creating a “full” MP3 file. The resulting file was 500 bytes larger. Why? Because of unnecessary header information that LAME likes to throw in just for neatness. This consisted of an empty ID3V1 header, padding for said, and a LAME parameters and version information tag. You could easily remove these and the file would still be a perfectly valid MP3. Just as streamable, just as nice.
In conclusion, just because you can shove an MP3 into a .wav, WHY DO IT? You gain a couple hundred bytes due to the missing ID3 and Paramater tags, but you lose the neat capabilities those tags provide, and you also lose streaming, not because the format doesn’t support it, but because nothing knows how to handle streaming WAVs, not knowing that the file is an MP3 until it opens it.
So now, can we agree on the following?
-Yes, you can create a .wav with an MP3 stream in it.
-Yes, the file is still streamable, its just that nothing tries to stream it.
-Yes, its a full MP3 file, it just has a WAV header wrapped around it.
Oh, and for your information, DivX;-)Audio is simply a hacked version of Microsoft’s Windows Media Audio codec, Version 7 as I recall, just as DivX;-)3.11Alpha was a hacked version of MS MPEG4 V2. If you like using unpopular codecs, I suggest OggVorbis. It has the best quality of any existing audio codec, and it’s COMPLETELY free. If you like the Windows Media sound (Vorbis is much better, try it), then use WMA8, or WMA9. Both have much better quality than DivX;-)Audio, and everyone can listen to them using Windows Media Player.
>> So now, can we agree on the following?
>> -Yes, you can create a .wav with an MP3 stream in it.
I do not know how to say this so that you will understand it: streaming formats repeat the encoding information periodically in the stream while WAV only has it once in the header. Take away the header of a WAV file and you have lost the decoding info and you have no way of correctly interpreting the rest. A streaming format like .MP3 repeats the same information periodically so that you can tune into a stream in mid stream and you still get the information necessary for decoding. I have said this repeatedly and you fail to grasp it.
>> Yes, the file is still streamable, its just that nothing tries to stream it.
NO! Take away the header of a WAV file and you cannot decode it. Just like if you take out the header of a JPG graphic you cannot decode the rest. But in MPEG video the decoding information is repeated every few frames so you can start decoding in mid stream even if you missed the beginning.
I challenge you (or anyone who agrees with you) to:
a) Find any cite supporting the notion that WAV files repeat the encoding information periodically and
b) find a program which will successfully decode a WAV file with the header missing or corrupted.
>> -Yes, its a full MP3 file, it just has a WAV header wrapped around it.
NO, it is most certainly not. Please show proof of that. You cannot stream a WAV file with a missing header. No way.
Regarding the DivX codec, I am very aware as we had a recent thread on that too. At any rate, a WAV file encoded with DivX according to you would be what exactly? A “DivX file wrapped in a WAV envelope”? There is no such thing as a DivX file type, only a codec, and there is no such thing as “wrapping” a different file type (.mp3, .au, etc) in a WAV file. A WAV file is a type onto itself.
At any rate, the proof is in the pudding: Please show me the same file, encoded with the same parameters, where the .MP3 is smaller than the .WAV. I say it is impossible and your own experiments so far support my position. Please provide a tangible example supporting your position.
Your entire concept that stuffing an MP3 into a WAV magically takes away the streaming ability of the MP3, and that somehow this is A) possible at all or B) reduces the file size is utterly wrong. I just proved this in my last post. Go make an MP3-WAV with your method, rename the file to MP3, and try streaming it. See? It streams properly. You haven’t changed the file, you’ve just changed the way its handled. Your MP3-WAVs won’t stream because WAV files can’t be streamed. The contents often can be, but the programs handling the files don’t know that. Do the exact same test YOU just mentioned. Take your MP3-WAV, chop some arbitrary amount of bytes off the beginning, and rename the file to MP3 so that the program handling it knows how to treat it. THE FILE PLAYS PERFECTLY FINE.
I have just conclusively and indisputably proven that your MP3-WAV files are simply valid .mp3 files with WAV RIFF headers wrapped around them. And no, I don’t think I’ll be spending hours with a hex editor accounting for and removing every bit of ancillary header information from an MP3 to account for 500 bytes of a 100KB file. The .wav filetype is simply a container that can be wrapped around any other type of audio, much like .avi is a container that can be wrapped around any kind of framebased video and any kind of CBR audio stream.
The DivX;-)Audio codec is simply a hacked version of Microsoft Windows Media Audio. As such, it is meant to go in a .wma file. Of course, no one uses it as such, and there are no tools to put it in WMA format.
Now, have I put this matter to rest? Or do you have further questions?
If I rename a .JPG file to .GIF my graphics program opens it just the same. I have just conclusively and indisputably proven that JPG and GIF are the same thing. ::sigh::
I guess we’ll just have to disagree and leave it at that.
Ok, I feel like jumping in the deep end here, and I’m going to have to agree with everything FDISK is saying regarding the wav vs mp3 - I had a quick look on google, but the search terms were just too general
so I did some quick practical experimentation, and encoded a 23 sec wav to mp3 and compressed wav @ 128Kbps, and lo and behold both were 372Kb (fancy that - 128 * 23 = 368, pretty close) but seeing as its small header info that we’re figting over I have to look at bytes
the wav came in at 372,034 bytes
the ID3less mp3 came in at 371,983 bytes
an increase of 51 bytes when using wav… so I open them both up in a hex editor. The mp3 goes into gibberish after about 7 bytes. the wav at 58. No, I’m not making this up. this leads me to believe that the data inside the wav I made with the included with sound recorder (MPEG layer-3, just above PCM) is infact a real honest to god mp3 stream with a RIFF header.
oh and who cares about the extention? Unix doesn’t, and IIRC mac doesn’t either, they look at the header of the file. Windows sends the file to a program which looks at the header of a file and decides what to do with it.