Theme by nostrich.
Post with 2 notes
This post was created as a response to a tweet regarding some audio editing accomplished with Melodyne editor; I had removed a horn sound from a speech sample and was quite happy about it . I took to Twitter and expressed my astonishment with Melodyne’s capabilities and Celemony requested I share audio examples (not sure how they intercepted the tweet in the first place?). What follows are Audio “before and after”s as well as a narrative of what I was attempting to accomplish.
I really think that Melodyne gets stereotyped - I’m sure that it’s a fine tool for fixing out of tune guitars and pitch challenged vocalists, but as a predominately Electronic musician I don’t do much of that - I like Melodyne for what it does “unnaturally ” to Audio.
Case in point: while working on a recent track of mine I decided that what was needed were a few “found” speech samples. I have to admit I’m a sucker for oddball speech samples (I grew up listening to Skinny Puppy, OK!), but I seldom end up keeping them in the final product (I always find them too “gimmicky”). When I do keep them, I mix them down low - almost at a subliminal level - which is what I decided to do on this project.
First task was to locate suitable material; after trawling some public domain film archives for related subject matter, I located an “ephemeral” film and extracted some relevant samples from the audio track for the intro and outro. However, the intended sample for the fade out was marred by quite a loud horn honk in the midsection:
Since I wanted to mix this down low and apply quite a bit of delay/reverb to the sample, the horn was too jarring - either the horn sound had to be removed or the sample was unusable.
I first tried applying some noise reduction (SoundSoap Pro is my favoured tool currently - at least as long as the now defunct Bias will continue to support it), and though it did clean-up the background noise a bit, the horn was still quite noticeable.
Editing out the sound was considered, but produced too much of a gap in the speech sample.
What about Melodyne? The horn was a high frequency spike in an otherwise low - voiced narrative which, I thought, should separate very nicely with the pitch detection algorithm…and sure enough it did.
After deleting the horn sound, some further editing with the amplitude attack and timing note length tools on the surrounding speech netted the following end product:
By no means perfect, but not meant to be - the material itself is poorly recorded from an old, “scratchy” film but fits quite nicely into the texture and the “Vibe” of the new track.
Thanks again Celemony for a great product…even if I don’t use it for its “intended” purpose!