least-evil ringtone format?

[miamicanes]

What's the least-evil format for waveform-based ringtones? I know that you can theoretically copy any valid media file to the right place and the phone will try to use it... but with both my Sprint Touch (latest rom) and old PPC-6700, it seems/ed like regardless of what I did, the act of using my own media file caused the phone to wait a few extra seconds before playing began (I'd see the screen power up and show an incoming call on the display, but would be unresponsive for a second or two, then would finally start playing the sound. Sometimes, it would stutter (ironically, seemed to be WORSE with uncompressed PCM .wav files than with MP3), and it usually seemed to cause memory leaks and lingering performance problems.

For what it's worth, the sound data files themselves were kept in system RAM... I was always careful to ensure they never went into the storage card.

Is this just an unfortunate artifact of ANY wave-based ringtones on the phones? Or is it a situation where the phone will TRY to play whatever you throw at it, but some encoding formats and/or parameters are "better" than others insofar as performance/memory/etc are concerned?

If all wave-based ringtones are problematic, what should I google for to find out what its "midi-ish" capabilities are? For example, I have a hunch that it's all software-generated and has few "hard" limits insofar as simultaneous voices, FM vs samples vs physical accoustic wave modeling, are concerned... but there's probably a point where it starts to impose more of a CPU load than an outright wavefile would.

If for some bizarre reason waveform playback is strangely demanding of resources, are there any extensions to allow Tracker-type modfiles to be played as ringtones?

[hibby50]

wavs will ring instantly but are huge. MP3s are smaller but are delayed. also keep them in my documents if possible, otherwise sd will do fine

[miamicanes]

Have you had any problem with PCM wavefiles stuttering, or otherwise seeming as if they were consuming disproportionate amounts of system resources when actually playing? The weirdest effect I remember seeing was that compressed files delayed, but seemed to have minimal impact on the rest of the UI or phone when playing... but playing wavfiles -- though faster to begin playing -- seemed to completely bring the whole system down on its knees.

I have a bit of a theory that might be on the right track, or totally off base:

* The phone might have limited amounts of bandwidth between system flash and system ram. Wavfiles don't need to be decompressed, but they involve more raw data to shovel between flash and ram.

* The phone is almost certainly using PIO instead of DMA to fetch data from flash into ram.

* It would (IMHO) be outrageous, but it's not inconceivable that it might have an onboard codec to do mp3 decompression in hardware, but directly bit-bang PWM via PIO. That would explain it perfectly, because THEN mp3 playback would involve feeding a low-bandwidth stream of data via PIO to the codec, vs having the CPU completely dominated trying to both shovel around large quantities of data AND intimately manage its playback (converting PCM to PWM isn't exotic, but can be computationally-intensive).

[scheel78]

I use mp3 about 30 seconds in length. I put the ringtones in windows\rings folder and I have no delay ringing.