
|
 |
Audio Production Hot Topics
We've arranged our audio tips into two categories: The Audio Production section has a few general tips for creating and compressing audio files, including tips on using WavToASF and the Microsoft NetShow T.A.G. Author, while
Third-Party Software gives tips on using Sound Forge 4.0, for example.
To record sound, you need, at a minimum:
Microphones to record with. If possible, use a quality microphone with a tight pick-up pattern.
A deck or other device to record into. Use a tape deck, a digital audio tape (DAT) recorder, or record straight into your computer.
For more detail, read the Recording Audio section of the "Working in NetShow: Audio" section of this Content Creation Authoring Guide.
To create digital audio files from your recorded sound, you need a PC or Macintosh A/V computer with:
A sound card to capture the sound. Use a 16-bit card with at least 2 MB of ROM. If recording to DAT, make sure your sound card
At least 16 MB of RAM to handle the large sound files. RAM is cheap; more is better. Windows NT® requires 32 MB.
Sound-editing software to produce and edit the audio files. A visual editor is preferable, and noise reduction and normalization features are a must.
For more specific suggestions, read the Transferring Audio to Your Computer section of the "Working in NetShow: Audio" of this Content Creation Authoring Guide.
Obviously a recording studio is best; it usually has good equipment and acoustics and is soundproof. If that's out of the question, you can still find or create a good place to record. You need a recording space that is free from high levels of ambient or background noise; watch out for loud computer fans or disk drives. For spoken-word recordings, use an acoustically neutral room without a lot of reverberation or echo. To test a room's echo, clap once and listen. Did that reverberate? Then spread carpets and hang drapes or blankets on the walls to deaden the echoes. It is relatively easy to create a makeshift studio, and it will drastically improve your recording and your compressed file.
The quick answer: It depends.
Different codecs work better for different types of audio. Some codecs work better with complex sound such as music. Others work best with less complex sounds such as a single moderated voice. Some codecs also work better with male voices rather than female voices.
So to speed your process of content creation, we tested the NetShow audio codecs with different types of content -- music, voice, voice and music, male, female -- to see which responded best in different situations. We concentrated on the lower bandwidths, since that's where it matters the most. Below, we have the following test results:
The .asf files were created in Sound Forge for an audio-only environment. Some of the codecs rated best sound quality (such as the GSM) may not work in lower-bandwidth illustrated audio and video files where they must share the pipe with other media. That's why we've also recommended a codec as "Best Bang for Your Baud" -- the codec which provides the best sound for the least bandwidth. Codecs whose quality was close to the winners we threw into the "Also Worth Trying" category. After all, this is subjective, and two different pieces of content may not respond the same way. So look at these as recommendations, not rules.
To give you a basis of comparison, you can listen to the files deemed Best Sound Quality, Best Bang for Your Baud, and Also Worth Trying.
Voice Only: Male

Vitals
Name: voice.wav
Duration: 21.29 seconds
Size: 459 KB
Sampling: 22 kHz 16-bit |
Summaries
11 kHz-12 kHz
Best sound quality: Voxware AC10
Best bang for the baud: Voxware AC10
Also worth trying: FHG MPEG-3 Layer-2, GSM 6.10
8 kHz
Best sound quality: GSM 6.10
Best bang for the baud: Voxware AC8
Also worth trying: L&H, DSP Truespeech |
Samples
Best sound quality: |
11 kHz |
8 kHz |
|
Best bang for the baud: |
11 kHz |
8 kHz |
|
Also worth trying: |
11 kHz
11 kHz |
8 kHz
8 kHz |
|
|
|
Voice/Music: Female
Vitals
Name: fem.wav
Duration: 21.4 seconds
Size: 833 KB
Sampling: 22 kHz 16-bit |
Summaries
11 kHz-12 kHz
Best sound quality: Voxware AC10
Best bang for the baud: Voxware AC10
Also worth trying: GSM 6.10
8 kHz
Best sound quality: GSM 6.10
Best bang for the baud: Voxware AC8
Also worth trying: DSP Truespeech |
Samples
Best sound quality: |
11 kHz |
8 kHz |
Best bang for the baud: |
11 kHz |
8 kHz |
Also worth trying: |
11 kHz |
8 kHz |
|
|
|
Voice/Music: Male
Vitals
File Name: male.wav
Duration: 11.37 seconds
Size: 491 KB
Sampling: 22 kHz, 16-bit |
Summaries
11 kHz-12 kHz
Best sound quality: Toss-up; FHG MPEG-3 Layer-2 or Voxware AC10
Best bang for the baud: Voxware AC10
Also worth trying: Don't bother.
8 kHz
Best sound quality: GSM 6.10
Best bang for the baud: Voxware AC8
Also worth trying: Um, never mind. |
Samples
Best sound quality:
|
11 kHz |
8 kHz |
Best bang for the baud: |
11 kHz |
8 kHz |
Also worth trying: |
- |
- |
|
|
|
Music: Male
Vitals
File Name: music.wav
Duration: 17.92 seconds
Size: 773 KB
Sampling: 22 kHz, 16-bit |
Summaries
11 kHz-12 kHz
Best sound quality: FHG MPEG-2 Layer-3
Best bang for the baud: Probably Voxware AC10
Also worth trying: Nope.
8 kHz
Best sound quality: GSM 6.10
Best bang for the baud: Voxware AC8
Also worth trying: Maybe FHG MPEG-2 Layer-3 |
Samples
Best sound quality:
|
11 kHz |
8 kHz |
Best bang for the baud: |
11 kHz |
8 kHz |
Also worth trying: |
- |
8 kHz |
|
|
|
It depends on the application you are using to create your .asf files. Though it is perfectly acceptable to create and edit files in any format you prefer, they must be converted to .wav format to work with NetShow's WavToASF and Microsoft NetShow T.A.G. Author applications. However, some third-party software, such as Sound Forge, can accept other formats and build directly to ASF.
NetShow Encoder can accept .wav and .avi files as an audio source.
"At what quality should I record/capture my audio?"
When possible, record and capture your sound at CD quality: 44 kHz, 16-bit. If, due to file size or equipment limitations, you must record or capture at reduced rates, lower your kilohertz (to 22, then 11) before sacrificing bit depth. The sound degradation is generally more noticeable between 16-bit and 8-bit sound than between 44 kHz and 11 kHz sound.
Always perform your edits on the highest-quality (preferably CD-quality) source files that you can record & capture, and then save a copy of this produced file before doing any downsampling or compressing. You will then have this high-quality edited file to recompress later when codecs improve and/or your bandwidth increases.
During downsampling, compromise sample rate before sample depth. The difference between 16-bit and 8-bit sound is more noticeable than that between 44 kHz and 11 kHz. If your data rate and codec will allow, stay at the 16-bit level. Note: The Lernout & Hauspie is a 16-bit codec; do not offer it 8-bit files. It will accept them, but you will have wasted bandwidth and unnecessary quality loss.
You can use any source. The better the quality to start with, the better the output will be. This is just like copying VCR tapes: If you make a copy from a copy from a copy the signal will deterioriate with each generation. The encoding process is a lossy process, so if you start with a poor quality signal and throw out chunks of the signal you will not have the same results as you would starting with a quality signal.
Microsoft NetShow T.A.G. Author Tips
If you are working with several audio files destined for the same ASF, you should join them into a single file before performing any edits. This will ensure that they join exactly as you wish and that they have uniform volume and quality. Always combine audio files into a single file before adding them to the Microsoft NetShow T.A.G. Author project.
You can only have one .aep file per directory. However, you can have more than one ASF per directory, with multiple files created from a single Microsoft NetShow T.A.G. Author Project. To create a second ASF, change the file name and build.
The Microsoft NetShow T.A.G. Author will create a default ASF filename based on the name of the Microsoft NetShow T.A.G. Author Project (AEP) project. To change this name (particularly important if you want to create multiple ASF versions from a single AEP), click File/Properties/Bit Rate Properties and enter your file name in the "Output File" box. Then click OK.
The Microsoft NetShow T.A.G. Author has a default end time of 1 minute. To extend this, you must first extend the timeline window beyond the desired end time. There are a couple ways to do this:
Import a sound file. This automatically extends the end time to the length of the file and the timeline to just a little longer than that.
To increase the timeline without importing a sound file, you must first click-and-drag the end of the timeline (in your editing window) beyond the point of your desired end time. Don't drag your arrow beyond the edge of the Microsoft NetShow T.A.G. Author window. If, while dragging, the timeline stops lengthening, just jiggle your mouse from side to side in the small space along the right edge of the window. It will get there eventually.
WavToASF Tips
The "-out" command is only necessary if you wish to specify an ASF file name different than that of your .wav file. Otherwise, WavToASF will automatically give your ASF the same file name as the source WAV.
-eccspan |
on|off |
Use this flag to turn error correction on or off. The default setting for error correction is "on." |
-in |
filename.wav |
Specify the input audio file that will be converted to an .asf file. |
-leadtime |
# of milliseconds |
Specify the maximum time before file playing begins. If there is packet loss, lower values may result in poorer audio smearing. The default is 4,000 milliseconds. |
-out |
filename.asf |
Specify a name for the output file. If you don't give the output file a name, WavToAsf uses the input file name as the name of the output file and then appends the .asf extension. |
-script |
Filename.txt |
If you want to add URLs, script commands, or markers to your output file, use this flag to specify the name of the script file to use. |
If you are using WavToASF and want to get a list of the WavToASF commands and their functions, just type in wavtoasf at the command prompt in your MS-DOS or Command window.
Third-Party Software
With your file open (and no portions of the file selected), choose Process/Resample from the menu bar to open the dialog box. Then enter your new sample rate (in hertz). Standard sample rates are 44,100, 22,050, 11,025, and 8,000.
Make sure that the box labeled "Set the sample rate only (do not resample)" is not checked. It is always good to create an undo and double-check that you've selected the entire file.
Finally, click File/Save As to save the file with a different name the source file, preferable something indicating the sample rate you've chosen, i.e. myvoice11.wav for an 11,025 Hz (11 kHz) file.
To convert to ASF, with your source file (see below) open, in the menu bar click File/Save As. In the "Save As Type" pull-down menu choose Advanced Streaming Format, in the "Format" pull-down menu choose your codec, and then click Save.
If your source file is at a higher sample rate than you want your ASF to be, probably. The only codec that doesn't require files to be at the target sample rate is the FHG MPEG-3 Layer-2. All others must be at the target sample rate before conversion.
Though the Voxware and Vivo codecs download for use with NetShow v3.0, they will not work in some third-party applications. To use the Voxware codecs, compress your audio file with the Microsoft NetShow T.A.G. Author or use NetShow Encoder.
|
|