I’m working on a mod for The Ur-Quan Masters (which incidentally is a free and awesome remake of the 3DO version of Star Control 2 that everyone should play) but unfortunately I seem to lack the necessary audio skills to pull off what I’m trying to do.
Essentially I want to record some additional lines of dialogue and, as seamlessly as possible, insert them into the existing game dialogue files in order to reintegrate information that existed in the original Star Control 2 PC release but is missing from the 3DO source code/sound files.
The problem (besides the fact that my voice isn’t the same as the original actors’) is that the original audio files have certain effects applied to them that I’ve been unable to re-create. I’ve tried using echo, gVerb etc at many different settings but I’m not having much luck.
I’ll be providing two original audio files - one for the Melnorme and one for the Mycon - as well as two ‘imitations’ I made of them.
What I’m hoping for is that someone could listen to the original files (called Melnorme.ogg and Mycon.ogg), then modify the imitiation files so they sound as close as possible to the original files (inflection of words doesn’t matter though). This will likely involve everything from modifying volume, to pitch, to adding effects, to saving in a lower quality audio format that matches the original (low quality) audio.
If you succeed, please let me know what you did so I can re-create it myself! Let me know what effects you used, and what settings you used for those effects. With that info, I should be able to create new lines of dialogue that I can more-or-less seamlessly fit into the originals, and will learn some more stuff about Audacity effects while I’m at it.
I’ve attached the original Melnorme file and the imitation file to this post. I’ll be posting up the Mycon files in the next post (due to the max 3 attachment limit). I would be extremely grateful for your help (and will gladly credit anyone who helps me get the audio as seamless as possible)
For the mycon effect drop the pitch by 10-20%, add lots of reverb,
the apply an effect called “thru-zero” from the mda pack, (or other flanger), with a rate of one second.
Then save with a sample-rate of 11025Hz and 8-bit depth.
Mix the original with a duplicate which has been pitch-shifted by about +50%, in Audacity you will have to use the time-shift tool so the pitch-shifted track is in-sync with the original.
Apply reverb to the original and pitch shifted tracks , maybe with a longer reverb time for the pitch-shifted track.
Then apply a comb filter with a frequency of 80-100Hz to both tracks
Then export (save) with a sample rate of 110250Hz and 8-bit depth (maybe try 4-bit depth).
[ The pitch-shifted track may need a phaser effect ].
Hm, not sure - the Mycon one sounds like it’s almost exactly the same save for some minor setting alterations, but the Melnorme one doesn’t really sound all that similar. Any other approaches that might work?
Oh, and do you remember the reverb settings you used for the Mycon one? Did you use any of the presets?
Alright, I’ve been playing around with the Mycon recording. I’ve recorded some of the ‘missing’ dialogue and tried to slot it in. See attached - the new part starts around the 10.898s mark.
I’m surprised how well it turned out, but it’s not exactly seamless yet. My voice is part of the problem (and probably not something I can do anything about), but the original parts also have some sort of ‘hiss’ or noise in the background, which my new part is missing. Any idea how to add it?
Also: how do I save an ogg file so it has the exact same settings as the original ogg file did? The only options I can see are a rather rough ‘quality slider’?
IMO matching the voice of another person so the audience can’t spot the join is impossible.
The only option for a “seamless” experience is to re-record all the dialogue.
The sizzly hiss does sound like a consequence of low-bit-depth.
You can emulate this by putting the code below into the nyquist prompt …
The lower the value the more sizzly the effect, ( try 32 instead of 8 ).
Wow, this turned out great! Thanks so much! The only issue was that the start and tail end of the recording (where there are moments of silence), the nyquist prompt effect kinda doesn’t work (it doesn’t add any hiss). So I used the Generate > Noise function to create some pink noise (at 0.1) at the silent points, then applied nyquist prompt effect over that to try and fix it, and it turned out rather good.
Probably about as seamless as I can hope to get it! Have a listen!
I guess next is the Melnorme. I haven’t had a chance to try your suggestion yet, but I will soon…
Hi Trebor
It might be a good idea to post code snippets as text. Graphics are not accessible, thus no chance to try your code if the user is visually impaired.
Thanks a lot for your great tips and insights into audio processing.
The consenus over at The Ur Quan Masters forums seems to be that the Mycon effects are pretty much spot on.
Unfortunately though I’m still struggling with the Melnorme effects. I’ve tried Trebor’s suggestions but so far my attempts all seem to lack that ‘richness’ and echo that are present in the original.
Does anyone have any other idea how to recreate the audio effects in the original Melnmore recording? My first post has the original sound file showing what I’m after (and another sound file with my imititation of the lines to give an effect-less version to play with).
…