Sure-Fire Tips for Encoding High-Quality, Low-Bandwidth Audio, Part 1
In the following tutorial, we’re going to explain how to get the best results when encoding audio for low-bandwidths. By low bandwidths, we mean up to 20Kbps audio for mobile phone networks and dial-up 28.8Kbps and 56Kbps targets. Anyone with more bandwidth to spare might want to listen up, too, as any bits you can squeeze out of the stream and still produce the audio quality your audience desires will save you bucks on your bandwidth delivery costs.
Editor’s note: While the following tips, tricks and techniques have been tested in real-world settings, your results may vary slightly depending on your audio content. We suggest you use the following tutorial as a foundation with which to begin experimenting.
What you’ll need:
The faster and more powerful the CPU, the better, especially for TwinVQ, a codec notorious for hogging CPU resources. Minimum: a Pentium III.
An audio pre-processor for filtering and instantaneous peak level control when live recording. (Check out limiting and compression products by Orban and Midiman Delta. Prices range between $1,000 and $5,000.)
A field mixer for live recording, so you can adjust audio levels and do some primary mixing of audio tracks. (Expect to pay $500 or more. Check out mixers by Tascam and Mackie.)
A separate microphone. (Lavalier, boom and shotgun mics start at $100 and go up to several hundred dollars. Check out mics by Shure.)
A way to capture audio for live streaming, either a separate card or included in video card. (Expect to pay around $300 for a separate card. Check out sound cards by Digidesign and Midiman Delta.)
A good sound editor for equalizing, editing and normalizing the waveform. Prices range from $350 for Sonic Foundry’s two-track Sound Forge editor to $10,000 for Digidesign’s top-of-the-line ProTools. Other editors worth checking out include Syntrillium’s Cool Edit and Steinberg’s WaveLab. Many of these software sound editors have demo versions available (keep in mind, however, that some of the more professional features may be disabled in the demo version). If a particular demo version is not an option, check out some of the freeware apps downloadable from the Internet.
Check out:
www.downloads.com (type audio editor in the search field)
www.analogx.com
http://www.topsitelists.com/start/audiosoft/
http://209.211.248.205/software/vstplugs/p1.cfmTidbit: Capture cards usually come bundled with a software editor. Digidesign ToolBox includes capture card and ProTools software for $550.
Of course, one of the most important things you’re going to need are one or two encoders with several codecs to choose from, depending on your content type and the bit-rate targets you’re shooting for. Following is a list useful codecs for low bandwidth scenarios.
WMA (Windows Media Audio) This is Microsoft’s home-brew codec found in its Windows Media Encoder. It’s a free patent license. It’s noted for its quality low bandwidth stereo files.
Real Audio codecs are developed by RealNetworks and are aimed at 5Kbps to 96Kbps target rates. RealSystems licenses the Sony’s ATRAC3 codecs for 96Kbps up to 352Kbps targets for its encoder.
ACELP (Advanced Code Excited Linear Prediction) is used for low bit rate voice. This is good for vocal-only audio because it limits frequency response to the vocal range and uses more efficient coding based on predictive modeling that can’t be done on dynamic music material. Both RealProducer and WM Encoder include this codec for speech encoding.
QualComm PureVoice is used in Sorenson encoders for vocal only encoding.
QDesign is used in Sorenson encoders for low bandwidth music.
AAC (Advanced Audio Coding) is used in MPEG-4. AAC has historically been used in broadcasting (is built into MPEG-2, the system used to make DVD movies) and for download and play at the 128Kbps and up level, although it may be useful for low bandwidth targets as well. Some people believe it compresses better than MP3. Fraunhofer, Sony, AT&T and Dolby hold AAC patents.
TwinVQ (Transform-Domain Weighted Interleave Vector Quantization), also found in MPEG-4, is reported to have excellent quality at low bit rates, but it takes more CPU power to encode and play back than most codecs. TwinVQ was developed by the NTT Human Interface Laboratories in Japan.
MP3 is a popular codec for music. It is known to produce quality music in download and play scenarios. Sony, Panasonic, and Compaq have MP3 players for this purpose. MP3 is now also being used for streaming audio. There are some bootleg MP3 codecs, but encoders for professional use should have the MP3 codec licensed by Fraunhofer Institute, the main developer of the underlying technology for MP3. Sorenson includes Fraunhofer MP3 in its Squeeze encoder. Sorenson suggests MP3 for music with voice, preferably at the 56Kbps or higher targets for audio-only without video. MP3 isn’t entirely free, however. A major shortcoming is its exorbitant licensing fee, a cost that is passed on to the consumer. It’s been reported that it costs $15,000 flat-fee for anyone wanting to produce an MP3 encoder plus $2.50 per unit. For playback, it costs $15,000 dollars advance for a player plus .50 per unit.
Ogg Vorbis is an open-source, free alternative to MP3 that some claim has superior acoustical modeling to produce better quality in smaller file sizes.
Recording Tricks
Tip 1: If you’re recording your own audio tracks, set up a separate sound stage and isolate it from noise. Install a floating floor, if possible. At the very least, use wide spectrum acoustic foam (check out products by Sonex and Auralex Acoustics) to cover a third to half of walls. This will reduce echo commonly associated with sound bouncing off parallel walls. Go to www.auralex.com for a tutorial.
Tip 2: Remove noise sources. That means anything a microphone could conceivably pick up: the fan from an air conditioner, a computer, a disk drive head searching for or recording data. Place the CPU outside the sound field and keep the mouse and keyboard inside the sound field by running twisted pair cable between both and using a switch box (Gefen Systems makes these for a few hundred dollars). Or, if recording to tape, turn off the computer.
Tip 3: Do not use the mic on your camera, because it can pick up camera noise. Instead, invest in a separate microphone. Lavalier microphones are good for talking heads, boom mics are good for two people sitting across from each other during a conference, and shotgun mics are good for capturing sound at a distance.
Tip 4: Invest a few hundred dollars in a good professional-grade microphone. Try to avoid tiny condenser mics if possible. Their use can result in hiss and distortion, eliminating the chances of producing a clear sounding audio file once it goes through a codec.
Tip 5: Keep the mic as close to the source of audio as possible. The farther away the source is from the mic, the more the noise floor is raised and the more bits have to go to vagrant sounds.
Page Two: More Recording Tips >>