Comments

Howard-Vigorita wrote on 12/1/2022, 12:19 AM

Noticed the same thing myself rendering long wav files to mp3. I keep an old vp14 handy on an older machine for that. Just rendered a 3 hour 24-bit wav file with highest quality mp3 preset in 2 min 54 sec. Vp20 rendered the same clip with same settings on same machine in 5 min 55 sec. Resultant mp3s were the same exact size. Go figure.

But I use vp20 for more modern video like 4k hevc. It's able to run rings around vp14 on newer hardware vp14 won't even recognize.

noisywan wrote on 12/1/2022, 4:46 AM

Thanks to Vegas gods we can have multiple versions running on the same system. (Well, I had a tiny issue right after installing v13 over v20, Track EQ stopped working and switched to DEMO mode in v20 but I fixed it by reinstalling v20. I guess the install order should be old-to-new by version wise.)

I hope this higher audio render times is a trade off for having a better quality output, not due to a poorly written mp3 plugin code that needs optimizing.

 

pierre-k wrote on 12/2/2022, 10:29 AM

If you want quality, DO NOT use the internal MP3 rendering.

rraud wrote on 12/2/2022, 11:21 AM

I think MP3 audio files can be encoded in Vegas with Happy Otter Scripts (HOS), using the ffmpeg codec. Otherwise the MP3 encoder in Sound Forge 16 is very fast. As many Vegas users are aware, Sound Forge can be integrated into Vegas.

rraud, Magix forums moderator

Howard-Vigorita wrote on 12/2/2022, 2:50 PM

I don't have sf16 installed on the xeon I tested vp14 on but did the same render with Sony SF7: 1:09 save; plus 0:34 building peaks.

On my 11900k I do have vp20, sf7, and sf16 but not vp14. vp20: 3:24; sf7: 0:47 save, 0:22 peaks; sf16: 0:33 save, 0:24 peaks.

@pierre-k You might want to throw a low-pass filter on the vp20 track (track eq high shelf with gain lowered all the way) and set the cutoff to max 24db/oct (or higher if you have better fx) and manipulate the cutoff frequency till you get those 2 graphs to look more alike. That will give you an idea of what the app that made the 2nd graph is doing. Looks like they set the cutoff somewhat below 20 khz to get the response shown. Which is good dac practice for wav sources sampled at 44.1 or 48khz. But folks record wavs at higher sample rates specifically so they don't have to filter so low or too steeply.

pierre-k wrote on 12/2/2022, 7:24 PM

You might want to throw a low-pass filter on the vp20 track (track eq high shelf with gain lowered all the way) and set the cutoff to max 24db/oct (or higher if you have better fx) and manipulate the cutoff frequency till you get those 2 graphs to look more alike. That will give you an idea of what the app that made the 2nd graph is doing. Looks like they set the cutoff somewhat below 20 khz to get the response shown. Which is good practice for wav sources sampled at 44.1 or 48khz. But folks record wavs at higher sample rates specifically so they don't have to filter so low or too steeply.

Aimp3 uses the Lame codec. That's why it has better results. Now I tried Vokouder and it is the same as Aimp3.

It's a shame that Vegas doesn't have the option of choosing a codec for mp3.

rraud wrote on 12/3/2022, 10:51 AM

I have not encoded many MP3 files lately, but I used WinLAME rc3 a lot which can batch convert and add HP and LP filters in the presets script, among other parameters not available on most encoders. For instance, my mono podcast preset:

<preset name="192kbps Mono Podcast preset">
   <comment>
 192kbps - 44.1kHz, mono, Filters: 10k x 85hz.
   </comment>
   <value name="bitrateMode">0</value>
   <value name="bitrateCBR">192</value>
   <value name="sampleRate">44100</value>
   <value name="hideLameSettings">0</value>
   <value name="quality">0</value>
   <value name="channelMode">3</value>
   <value name="highpassUse">1</value>
   <value name="highpassFreq">85</value>
   <value name="lowpassUse">1</value>
   <value name="lowpassFreq">10000</value>
  </preset>

That said, my old and abused ears could not hear any difference between the WinLAME and Sound Forge's Fraunhofer encoder using a HP/LP filter plug in SF. The WinLAME encoder was relatively fast as well. FWIW, encoding in mono is the resolution equivalent of a 384kps stereo MP3. Of course any spacial information would be lost, but there typically is not any on voice recordings.