3gp...

[raymond904]

Can anyone help me with converting a 3gp video? Ok I get that the files are supposed to be small - but I'd like to be able to make out the audio.

I am using Xilisoft video converter and I can up the video bitrate so that it looks good, unfortunately it won't allow me to up the audio sample rate from 8000 and so the sound is terrible no matter what bitrate I choose.

[erayboul]

I'm not that familiar with the 3GPP and 3GPP2 formats, but I did create a short 3GPP video from a DV-AVI with Quicktime Pro. As you probably know those two formats are primarily for video on cellphones. I just don't think that it is possible to make either the video or the audio any better than it already is. The information is simply not there to make it look or sound better. For example with a 8000 bps sampling rate for audio the frequency response is 0 to 4000 Hz, which should be OK for speech, but there is no treble (high frequency) information. It is impossible to get that treble information if it was never stored in the encoded file. Similar principles exist for the video information. 3GPP and 3GPP2 are extremely lossy compression schemes. Once the data is gone there's no getting it back.

[Cornucopia]

Pre-filtering (with high quality dithering) can help somewhat.

Audio Example:

Source Audio = 16bit, stereo, 48kHz, LPCM ?
0. Mix to Mono.
1. Use Dyn. Range Compressor to get levels w/in ~48dB range, w/ Limiting of major peaks.
2. Use Harmonic Synthesizer to octave double lowest bass ranges (20Hz-80Hz --> 80-320Hz (added/mixed at low level to main signal) (only if really needed)
3. Use Peak EQ filter of ~+3.5db for ~1.2kHz, 1.5octave band
4. Use LowPass(HiCut) filter (w/ high phase linearity--FIR) set to 3950Hz
5. Use HiPass(LowCut) filter (also high PL--FIR) set to 80Hz
6. Use a dithered Normalizing Limiter again to punch everything up to the top of the dyn. range.
7. Apply "Noise Reduction/Smoothing" filter to pre-subtract elements that can be eliminated prior to compression--This is a good way to hear an output similar to what the psychoacoustic masking will be doing in the compression. (use sparingly and with caution)
8. RENDER elements of sound/music/speech inactivity AS digital silence (all zeros). This can help alot!

Now, compress. Use VBR/ABR if you have the option and the playback devices support it. Use the highest quality encoder available.
.
.
.
Similar gains can be done in video as well...(that's for another time).

Hope that makes a difference for you...

Scott

>>>>>
edit: Actually "similar loss savings". The idea is: You are directing the loss as opposed to letting the compressor arbitrarily determine the loss, thus sonicly maximizing the available remaining bits.

[madav]

try using 3GP Converter 2006 the output quality is really good.

[lantern]

You could try SUPER to encode. It works pretty good for 3gp file creation.

[reboot]

Another freebie is the Nokia mulitmedia converter. It's a free download.
ImToo also has a 3gp converter that looks identical to the Xilisoft one.