Some more info:
AVI is not a codec, but a container format for video and audio streams, both usually compressed by codecs. For codecs you have a number of choices that can be used for video & audio – which to pick often depends on where the final video will end up for viewing, and which codecs you have on your system.
FLV is a proprietary format, Flash video, which is in essence both a video codec and a container format. AFAIK you can’t do FLV into an AVI container.
I’ve had some success using 2.57 (no a or b) to run some tests of the FLV option, both silent and muxed with audio. Using the raw PNG image sequence output for my movie Kata, I’ve been able to produce good-quality FLV-format files with the AAC audio codec that are comparable in size to AVI with Xvid & mp3 compression. Personally I prefer AVI/Xvid’s quality for the video but the FLV is acceptable when that format is required.
The biggest questions regarding using these options are about the settings for bitrate and GOP size, and similar settings for the audio codecs. There are no defaults in Blender except some that are used in the presets, so experimentation is in order.
Just FYI, my FLV test outputs produced the best results with the Kata footage using these specs:
Output – MPEG (required for audio/video multiplexing, aka “muxing,” in Blender)
Encoding – Format: Flash; Bitrate = 3000 ; GOP Size: 18-72 depending on the subject matter – generally, lower numbers = better quality = larger file size; Rate: Min = 0, Max = 9000, Buffer = 1792; Mux: Rate = 1008000, Packet Size = 2048; Audio Codec = AAC; Bitrate = 224, Sample Rate = 48000, Volume = 1.0
I have to admit some of these specs are opaque to me, I basically copied them from one of the presets to start with, but they seem to work OK. My PNG image sequence was loaded into the VSE along with the pre-mixed audio track in WAV format. My image size was only 768 x 432, so the VSE ran it out pretty fast, about 0.15sec/frame for 7160 frames. YMMV.