I'm trying to create a "vocal pad", just to clarify an ethereal voice singing just a single, sustained vowel. I'm approaching by using many band pass filters ( ie formants ) trying to emulate human voice, but the result is poor. Just asking if a good result can be obtained in this way, or in some other, or if I need samples of a real voice.
Are you automating those band pass filters? The filters must move over time to get voice sounds.
What software are you using? I achieve vowel like sounds in ableton by taking a synth and running it through two parallel chains, each with single band pass on them. One cut off should be in the low freqs like around 350 hz, and the other in the high freqs around 12khz. Then I map each cut off to the same knob with inverted mapping and ranges that range from their start values to near middle.
So when the control knob is at position 0/100, both band pass are at their respective starting freq (mentioned above at 350hz and 12khz). When the control knob is at 100/100 both bandpass are near mid values, maybe like low = 1khz and high 2.5khz.
You have two play with different values, but the key is as follows.
- Two bandpass filter
- one set high, one set low
- move in opposite directions
Then you take this effect chain and drop it on another chain so this filter effect is only applied about 50% wet.
If you pick a well developed "physical modelling" synth such as the following:
Then it sounds like you can get what you're looking for. Just don't use a crap implementation but a good one or try to do it yourself if you don't know what you're doing (don't try to replicate years and years of university research).
The "choir" effect is done by using multiple of such "single voices" in chorus.
The Delay Lama
is a classical plug-in in this genre, but I've seen some others as well such as
Also using a high quality real vocal sound in Paul Stretch or similar might be useful for "vocal pads".