Additionally, if it's just voice(or simple tones) you are storing, you could encode them as speex or low quality mp3's before storing them with Base64. It would save alot of text space. Of course, your end users would need the decoding software on their end, but most people have mp3 decoding ability. You also could use something like Sox to convert your high quality stereo wavs, down to low-bitrate single channel wavs, to save space. For instance, this audio link is only 9.6k of binary audio, and is 13k after being base64 encoded.