Chipmunk voices in second audio stream

[zambelli]

I'm trying to add secondary audio stream to a DivX movie. Both audio streams are MP3 (VBR) encoded. Since the second audio stream is director's commentary, I would like to use a very low bitrate in order to save disk space (it's only voice, after all, it doesn't need 320 kbps :-)).

I'm using Lame 3.92 to encode the MP3 files and Nandub 1.0rc2 to mux them together with the video. I'm using the MMSwitch DS filter to play them back in Media Player 6.4.

The problem is the following: when I encode the secondary audio at 32 kbps or 40 kbps, Lame resamples the audio to 22kHz. When I mux the audio in Nandub and play the final video file, the primary audio is just fine but the secondary audio stream plays at a high pitch and makes everyone sound like the chipmunks.

I am assuming this happens because the secondary stream is encoded at 22kHz while the primary one is encoded at full 48 kHz. But if I force 48 kHz on the low-bitrate MP3, it will sound horrible (you can imagine that spreading a 48 kHz spectrum over a 32 kbps bandwidth is not a good idea).

Is there any way to get this to sound good while still using a low bitrate (40 kbps or less)?

[tiki4]

Ever tried Ogg Vorbis? I guess the Ogg container gives you that kind of flexibility to do what you like to do.

Try the first soundtrack with quality -q 0.4 in BeSweet and the commentary track with -q 0. The first should give something like 130 kBit and the second around 64 kBit. Then mux with your video with Koepi's OggMux. You need the OggDS from Tobias Waldvogel as well. However there are excellent guides about Ogg and Ogg Vorbis on doom's site.

Regards,

tiki4

[Alestrix]

Originally posted by zambelli
The problem is the following: when I encode the secondary audio at 32 kbps or 40 kbps, Lame resamples the audio to 22kHz. When I mux the audio in Nandub and play the final video file, the primary audio is just fine but the secondary audio stream plays at a high pitch and makes everyone sound like the chipmunks.
[...]
Is there any way to get this to sound good while still using a low bitrate (40 kbps or less)?
There has been discussion about this in several threads already, but here's one way to solve it:
Encode the secondary Audiostream as low bitrate CBR-MP3 (not VBR/ABR) and add a WAV-header with a tool (I use wavemp3.exe, but don't remember where I d/loaded it from, BeSweet might be able to do it, too.). Then you can mux that "wav" in with NanDub. At least with VDub it works this way. If NanDub doesn't want to make it werk, mux the WAV with VDub first, open the file in NanDub and chose "Direct Stream Copy" for the primary audio stream and mux in the main audio as secondary (not the most comfortable thing at playback since you have to switch audio tracks at the start, but it works).
The suggestion with OGG Vorbis is feasable only if you don't mind to reencode the primary audio to Vorbis as well and mux with OggMux, since NanDub (at least on my computer) crashes on the attempt of muxing Vorbis into an AVI.

Hope it helped...

- A

[tiki4]

No, not Ogg Vorbis in AVI but Ogg Vorbis in Ogg container!

If you don't like to reencode your movie soundtrack as Vorbis, then how about muxing the MP3 into the AVI with the movie first, then you open your AVI file in Graphedit (Render media file), remove all filters after the AVI splitter, connect both audio and video stream to the Ogg multiplexer of Tobias Waldvogel and also the Vorbis file of your commentary track. The output pin of the multiplexer has then to be connected to the File writer filter. That's it. There are excellent guides on how to do that on this page and you avoid all problems of VBR MP3 in AVI containers.

Regards,

tiki4

[DJ Bobo]

If your mp3 is mono, MMSwitch will screw it up.
Use BSPlayer to play and all will be fine.

And another point: if you see MPEG2 or MPEG2.5 when encoding to MP3, your MP3 will be probably asynch, so be sure to see MPEG1 while encoding.