another GPS idea - speed-sensitive voicemail

Shachar Shemesh shachar at
Sun Jul 29 12:25:10 CEST 2007

Ian Stirling wrote:
> There are 2 D/A, 2 A/D flexibly routed, and one D/A that is dedicated
> to the earpiece.
> About the only limitation is that you can't do things that would
> require more IO sources than are available.
> For example, playing stereo MP3, and acting as voicemail/answerphone
> may not be possible.
> (It'd have to drop to mono).
Lost you there. You seem to suggest the following route for recording
voice calls:
1. Call arrives compressed with a GSM codec
2. Phone decompresses codec
3. Phone moves uncompressed stream through D2A
4. Phone further moves stream through A2D
5. Phone compresses the resulting stream
6. Phone saves compressed stream, presumably to the flash

Why not just do:
1. Call arrives compressed with a GSM codec
2. Phone saves compressed stream to flash

I really don't see why the A/D infrastructure needs to be involved in
voice recording at all. In fact, it seems that it should be easier for
the phone to save the call than to play it to the speaker.


Asterisk, for example, saves most of its recordings (pick up greeting,
extension selection, voice mail greeting etc.) saved while compressed
with GSM codec. As far as I understand things, if OpenMoko did that,
playing a recording would involve getting it off the flash and dump it
into the GSM line. Extremely light on CPU, and thus unintrusive.


More information about the community mailing list