Thread: Producing Audio for 3 axis skulls and VSA routines

Tweet

I have been asked to do a tutorial about the production of the tracks that I use to create the routines for my Skeleton Quartet.

I will start here and add to it as I go along. This tutorial will cover:

• The order in which to do things
• Software to use
• What to do if you can't sing
• Where to get the music
• Where to find talent for vocals
• and anything else I can think of as I go along

But we should start at the beginning. First, we need to identify what you are programming for. What I will talk about will work for all VSA routines whether you have a single skull, a quartet like mine or any other configuration. The only thing that changes is how many tracks you have which is determined by how many skulls.

For purposes of this tutorial, I will assume that you have a quartet like mine shown here:
Singing Skeleton Quartet Videos 2009-2011

The very first thing you need is an idea.

You have to decide if you are going to program just dialog, or singing or both. Then if you choose singing, you need to decide if you will use an existing song or if you want to make one yourself. This part all depends on your own creativity. You can create a totally original work, or like I do, write a parody.

I prefer to do parodies because the tune is recognized by the audience and if you use clever plays on words or themes of the songs, a parody can be very humorous. I also do parodies because I am not clever enough to come up with an original composition of my own.

So let's proceed with the parody concept. You have just heard a song on the radio or a tune has been stuck in your head and you think of a funny line that has something to do with Halloween or skeletons, or zombies or anything related that would be a funny twist on the original song. Write that thought down.

Then use Google to obtain the actual lyrics for the original song. There are hundreds of sites that can provide that.

Go to iTunes or wherever you purchase your music and download a copy of the original song that you wish to parody if you don't already have it. It is very helpful to listen to the original song over and over in detail to get a feel for how the original artist performs the song, what phrasing they use, how they emphasize certain words and anything else that can help you in the performance of the song.

Your next step will be to determine what to do about the instrumental track, assuming that the song you chose was not sung A-Capella. I have done some of those, and they are tricky, but you don't need a music track and they sound great if done well.

For these purposes, we will assume that the music track is distinctive and will tip off the audience about what song you are going to parody as soon as they hear the music start.

You have a few choices here as to where to get a music background track.

The easiest solution is to purchase a Karaoke track. You can buy these on iTunes as well. just add the work karaoke after the title of the song and choose the result that says, "Sung in the style of" whatever artist you are parodying.

There are other solutions as well. If you are talented in this are or know people that are, you can play the instruments yourself and create your own sound track. Some of my songs use this method, although I do not play the instruments, I have met someone who can and does do some tracks for me.

Another solution is to create the music using music software. I have done this with several songs. An example of one of these songs that I created completely with computer software is the track from my "Jingle Bones" routine. You can see it here:
http://www.youtube.com/watch?v=YfGBfgKPQXc

This is an early routine I did before adding arm motion to the quartet. It was one of my Halloween Carol series where I did Parodies of Christmas Carols. A member of this forum even wrote the last verse when I posted here asking for help.

Anyway, I wanted a version of Jingle Bells that was in a minor key to better suit it for Halloween, and couldn't find a suitable instrumental track anywhere.

I purchased a program called "Fruity Loops" from Image-Line (http://flstudio.image-line.com/fruit...direction.html) and created the music track that you hear by itself here: http://halloweenBobsHaunt.com/JingleBonesMusic.mp3

It helps to buy a copy of the sheet music for a song if you are going to attempt to create your own music track, or even to help with singing the correct notes and harmony parts.

You can purchase sheet music for just about any song from one of these online vendors:

http://www.musicnotes.com
http://www.sheetmusicplus.com
http://www.sheetmusicdirect.us/?refer=smduk&country=US

And many others that you can find. Most will allow you to download the music you want instantly. Others will make you purchase the music and will mail you hard copies.

You may have to transpose the song if you want to put it in a minor key like I did and it's helpful if you know something about music and know how to read it if you are going to attempt this, but having the sheet music is always a good guide.

Even if you don't know much about it, it is still possible by trial and error to move around notes within the software until it sounds the way you want it.

Let's say that you play guitar or piano, but that's it. You can still play some of the parts yourself and just add the missing tracks from a great place on the web that allows you to customize the karaoke track you want. look here for an example:

http://www.karaoke-version.com/custo...lee/fever.html

This is great. I was working on a brand new parody of Peggy Lee's "Fever" for my quartet. I am currently still working on it. For the music track, the real song uses only a bass guitar, drums and finger snapping. I decided that for my version, I would replace the Bass Guitar with a Bass voice that one of my skeletons would sing. (I did the vocal) I had fingers and knew how to use them, so I also created the finger snapping track. Then using that website, I selected ONLY the drum track and downloaded it, then added it to my mix. It came out great. I just have to finish all the other vocals and make something up for the other two guys in my quartet and Mr. Head to sing.

I should add here, that the goal is to get each voice on a separate audio track. If you have 5 singing skulls, then you need 5 separate audio tracks, each containing only what that particular skull will say or sing. This is so that you can program the jaw movement using that individual audio track in VSA for each skull. The music and any other sound effects must be on a separate track. All these tracks will get mixed together for the final product which is what your audience will hear when you are finished. I will go into much more detail about how to do the Jaw Programming and all other programming in VSA later in this thread. I already have a tutorial on this topic here: http://graveyardskulls.com/tutorials...mming-101.html

So I will wrap up this first installation of this tutorial and leave you in the following position:

1. You have had an idea for a new parody song
2. You have downloaded the original song mp3 and the lyrics.
3. You have written out all your new parody lyrics.
4. You have purchased, created, or conned someone else into creating your music track
5. You are now ready to sing your new vocals.

That is where we will pick up next time.

I will list here all the software that I use regularly to create the music tracks that I make. I will try not to forget anything and will address how and when to use all of these throughout this thread.

1. Fruity Loops http://flstudio.image-line.com/fruit...direction.html
2. Adobe Audition 3 or higher http://adobe.com (adobe.com probably no longer sells version 3, but you can get it for a good price on ebay and other places.)\
3. Sony Sound Forge http://www.sonycreativesoftware.com/soundforgesoftware
4. MIDI Converter http://download.cnet.com/MIDI-Conver...-10506124.html
5. Melodyne studio by Celemony http://www.celemony.com/cms/index.ph...roducts_studio
6. VSA (Hobbiest) by Brookshire Software http://www.brookshiresoftware.com/vs...7mvtfnfm9d6pl4
7. VSA Console (Sold at the same website as VSA.
8. Virtural Haunt by Monkeybasic http://www.monkeybasic.com/Products/VirtualHaunt/
9. Helmsman by MonkeyBasic http://www.monkeybasic.com/Products/Helmsman/

I'm sure there's more, but these are mostly the essentials that I use all the time. I hope this produces some helpful info for you.
If you are on a tight budget, and who isn't, skip down a couple posts and I will present some alternatives to the somewhat pricey software I have listed above and will list later.

OK, today's post is all about how to produce your individual voice tracks and why it is important to do this.

To do the recording, you will need to have several things ready to go. I will list them here. You may not need every one, but I will explain what they each are for.

1. A Decent Microphone. (I recommend the Sampson CO1U USB Condensor Microphone. You can get them many places including eBay. Here is one link: http://www.bhphotovideo.com/bnh/cont...=REG&A=details
2. Adobe Audition 3 or higher http://adobe.com (adobe.com probably no longer sells version 3, but you can get it for a good price on ebay and other places.)
3. Sony Sound Forge http://www.sonycreativesoftware.com/soundforgesoftware
4. MIDI Converter http://download.cnet.com/MIDI-Conver...-10506124.html
5. Melodyne studio by Celemony http://www.celemony.com/cms/index.ph...roducts_studio
6. Winamp or Windows Media Player
7. SmartScore X Midi Edition from Musitek http://www.musitek.com/midiedition.html
8. A quiet room to record in.

I use that microphone because it is relatively inexpensive and has a really good professional 'studio' sound to it. When I ask other people to record parts for me, I buy more microphones and ship them to those people. That way, all the tracks have the same dynamics and sound like they were recorded in the same place on the same equipment.

The next step depends on you. It is time to record your first vocal part. If you are already a good singer, you may not need to create a 'practice track' to make sure you sing all the right notes. I do need such a track. So I will present a few ways to create one.

One method I use, especially if the parts I am going to record are close harmonies is to create an audio file with just the notes I am going to sing played out on a piano. Then I play that track back in winamp or Windows Media player into my headphones while I sing into the microphone and record it in SoundForge (Or any other recording software) right on the same computer. Since SoundForge (or whatever software you decide to use) is only 'listening' to the microphone input, you end up with a recording of only your voice.

To generate the piano track you can do several things. You can play it on a piano (or any instrument) and record it, but this may create problems as to exact timing when you go to put all your parts together. What I do is to use SmartScore and the Sheet music that I purchased. Within SmartScore, you can scan the sheet music and it will create a MIDI file for you. Before you create the file, you can edit the score and remove all notes except for the ones related to the specific part you want to create a singing guide for.

I will use examples from a song I did called 'Rise Up Ye Buried Gentlemen'.

I got the sheet music for "God Rest Ye Merry Gentlemen" and scanned it into SmartScore. If I played it and created a MIDI file it would sound like the following short snippet:

http://HalloweenBobsHaunt.com/4PartsGentlemen.mp3

By editing out all the notes except for the one line I wanted, for the same short snippet for the lead part, I get this:

http://HalloweenBobsHaunt.com/Buried...1Line1Midi.MID

When I go back and re-edit eliminating everything except the second part I get this:

http://HalloweenBobsHaunt.com/Buried...3Line1Midi.MID

The third and fourth parts are as follows:

http://HalloweenBobsHaunt.com/Buried...3Line1Midi.MID
http://HalloweenBobsHaunt.com/Buried...4Line1Midi.MID

Now you know what it all sounds like together and you know what each part sounds like individually. I would break the song up into short segments like the ones I just posted and play them back in WinAmp to my headphones in a continuous loop. Then after hearing that short part go by several times in my ears I would sing along with the recorded piano, singing the section of my new lyrics that I wrote into the microphone over and over again for several takes. This way, I also get the timing correct if I keep time with what I hear in my headphones. If I am off in some places (Which I always am) it can be corrected later. If you can't sing on key (Like me) don't worry, just be confident and come close. Pick a note and stick to it. Don't let your pitch waver or correct itself, even if it's the wrong note. You can fix the notes later, but it's hard to pitch correct a note where the pitch changes throughout that one note.

Once you get several takes of each part for the same phrase, you should end up with something like these parts:

http://HalloweenBobsHaunt.com/HF/Part01.mp3
http://HalloweenBobsHaunt.com/HF/Part02.mp3
http://HalloweenBobsHaunt.com/HF/Part03.mp3
http://HalloweenBobsHaunt.com/HF/Part04.mp3

They are close after several takes, but still not timed out perfectly and still have some sour notes.

Here's what they sound like when put together:

http://HalloweenBobsHaunt.com/HF/Gentlemen4Parts.mp3


Now it's time to fix things up.

I can not go into detail on how to use all the features of each software program within this tutorial, but there are plenty of guides and tutorials for each piece of software both on the manufacturer's websites as well as on YouTube.

The first thing we have to do is to use a multi-track editor to line up all the syllables of all the words and make sure that they all sync together so that the lyrics can be understood.

So open Adobe Audition (or get Audacity for free http://audacity.sourceforge.net. Audacity is also a multi-track recording and mixing software, but has limited features compared to Adobe Audition)

Drop each of the tracks into Audition (or Audacity) and set them up one part for each track. You can use this software to align small parts of each track directly in line with the lead vocal until all the syllables line up. If you were short on one syllable, you can even time stretch it out to fill the space so it matches the lead vocal line that you wanted to line everything up with. Even if you end up with a choppy sounding individual track or two, it won't be noticed once you drop the whole thing in the final mix. You can experiment by using the files I provided.

Once you start working with Adobe Audition or Audacity, always save everything as a WAV file. Each time you save a file as an mp3, there is a small space added at the beginning and end of the file which is translated as a short silence. This is where the compression information is stored and since it is not audio material it has no sound. It does push the audio out a bit and will start to unsync all your work. WAV files remain as you created them. I only use mp3s here because they have a smaller filesize and download quicker.

Now, once you have all 4 files spaced out properly so all the words and syllables line up, you need to pitch correct them unless you are a good singer and hit all the right notes already. If that is the case, then you are basically done here and don't need to use the pitch-correction software at all. I wish that were the case for me.

After aligning them, they will sound like this when all played together:

http://HalloweenBobsHaunt.com/HF/Gen...PartsTimed.mp3

And here are the time corrected 4 individual parts (Still with the sour notes):

http://HalloweenBobsHaunt.com/HF/Part01Timed.mp3
http://HalloweenBobsHaunt.com/HF/Part02Timed.mp3
http://HalloweenBobsHaunt.com/HF/Part03Timed.mp3
http://HalloweenBobsHaunt.com/HF/Part04Timed.mp3


Now we have to pitch correct.


First, open up Melodyne studio by Celemony. (This program will only work with WAV files, so make sure all your files have been converted or saved to WAVs)

Go to 'File', and then 'New arrangement'.

Then click on the box for 'Track 1'. Next go to 'File', then 'Import Audio File'

After converting or saving all your files to WAVs, including the MIDIs of the piano tracks that you used as singing guides, you can navigate to those files. Open the piano track (the converted WAV file that you made with your MIDI Converter) and import it to Track 1.

You should make sure after creating the WAV file from the MIDI that the notes line up with the timing of the words you sang. They should all line up as soon as you get it placed in the right starting position. You can line these up in Audition by using your vocal tracks as a timing guide for the original piano tracks.

OK, so you have your aligned Singing guide piano track WAV file in track one.

Click on Track 2, go to 'File', then to 'Import Audio File' and select your voice track for the same part as the singing guide and place it in track 2.
Now double click on any note that you see in track 1. It will open a new window.
Go back to the original window and double click on any note from track 2. It will add that track to the new window.

If you made no errors and were in perfect pitch all the way through, the notes should be sitting right on top of each other. Perhaps they will be an octave different, but will be the same notes, for example, both A#'s but an octave apart.

Click on a note from your vocal track and make sure that is the track highlighted. Now hold down the CTL key and click A. This will select ALL the notes in your vocal track.

While all the notes are highlighted and selected, hit CTRL+ALT+P all at the same time. This will align all your notes to the closest 'real' note. For example, if you were just a little sharp or flat on a note and it was in-between two notes on the scale, doing that will move your note to the closest real note on the scale from where it was.

Now all you have to do is to drag each of your notes so that they line up exactly on top of (or on the corresponding note an octave apart) the notes in the singing guide.
Once you have made all your adjustments, close that new window, and go to 'File', then 'Save Audio'.

Select 'Save Each Track as a single File'.
Select the vocal tracks by checking off the boxes in front of them. You can have them all loaded in here at once if you want. You will have to learn this software.
Name the file (or folder if there are multiple files) and your new WAV files will be created with you singing in perfect pitch.

Here are the resulting files created by altering the same earlier files we have been working with throughout this post.

The four separate files:

http://HalloweenBobsHaunt.com/HF/Par...AndPitched.mp3
http://HalloweenBobsHaunt.com/HF/Par...AndPitched.mp3
http://HalloweenBobsHaunt.com/HF/Par...AndPitched.mp3
http://HalloweenBobsHaunt.com/HF/Par...AndPitched.mp3

And now the mix of all the files together:

http://HalloweenBobsHaunt.com/HF/4PartsPitched.mp3


There. Simple.

I started with a terrible mix and was able to correct all the timing and pitch errors and end up sounding like a professional Barber Shop Quartet, and you can too.

I am happy to help anyone who is interested to use the software if you get stuck along the way, and I will, of course, answer any questions you have.

The one thing I did not mention here is that the final mix in the video shown here: http://www.youtube.com/watch?v=3YDeOc6COac
has a music track to it.

I found this music track online and just added the intro from toccata and fugue in d minor to it. The first thing I did was to adjust everything so that the tempo I was singing matched the music track. I built my mix by adding each voice track to the music track and adjusting the timing to match the music track.

I left that step out here because I was focusing on creating the vocal tracks. A song like this could just as well have been A-Capella also.


Any questions?

I realize that I have listed a lot of high end software with price tags to match. I do this because this is the software that I use and am most familiar with. I use all this software because I have been in Radio for many years and have obtained this software by purchasing it myself for my business or in some cases it was purchased for me by my employers.

If you don't have those options, there are many ways around this. For example, I list Sony Sound Forge as the software for capturing your audio from the microphone. Sound Forge has many great features but if all you are going to use it for is to capture your audio, then don't buy it. Adobe Audition can also capture your audio. So can hundreds of other freeware solutions that can be found on the net. In fact, windows sound recorder can even do it. Just find something that can save as a WAV file.

As I mentioned before, Adobe Audition is also wonderful, but Audacity is a freeware alternative. It has fewer features and I am not completely familiar with it, but it does do all the major functions that Adobe Audition can do. Maybe it means that you will need to sing more takes to get the part closer to the exact timing so you don't have to modify it as much, but maybe that's worth the extra savings.

The vocal pitch adjustment software I mentioned is not needed if you sing on key to begin with. I don't know of any freeware solutions for this, but there may be some out there.

Also, you can save big bucks and get the great software I am using by purchasing slightly older versions of each product at places like eBay. Go for Adobe Audition 3 or 2 instead of 5 and maybe you can get a copy for $50.00.

Look for an old version of Sound forge 9 instead of 10 or even sound Forge 7.

That will save on the budget as well.

Thanks for sharing this information. I think very few understand all that must go into producing just a few minutes of these routines or skits. Looking forward to watching this thread and at some point, hopefully, of contributing. As for now, I'm all ears.

Great information and really nice work. I've been milling around an idea using a jaw-synced 3 axis skull and some other props singing "Monster Mash." My question is how do you sync all the "actors" for the final song? If all the jaw movements are the same for each singer it would seem pretty easy, but what about four singers with four different timings? I'm imagining actually recording each final track on a sound chip and syncing the start of each chip (along with all the choreography) with one master microcontroller. Thoughts or ideas?

bkr1969,

You sync all the 'actors' using VSA software. You program in all the movements, and just play it back. VSA keeps everything synced up to the music track.

The thing you will need to do differently about creating vocal tracks is that you don't need to sing in key. You don't even need to say all the right words. You do need to have the timing down.

What I do to create a vocal track that I will use to program a jaw movement in VSA, is to listen to the real song in a set of headphones. Then I sing the various parts into my mic and record them on my computer. Monster Mash has a lead vocal and several other voices in the chorus. There is also Dracula saying "Vhat ever Happened to the Transylvania Twist?". You can direct that line to a different skull that is supposed to be the good Count.

So you decide what parts of the song will be sung by which skulls. Play the song in your headphones and sing along. Once for each skull. Once you have created a vocal track for each skull, use your multi-track software to make sure that all your words and syllables sync up as close as possible to the original song.

Once you get those put together, your ready to program. It doesn't matter what you sound like since those tracks will never be heard by the public.

Using VSA's 'Wavemotion Analysis' feature, you create the jaw movement to each of your audio tracks.

Then, program in all the head movements and it's done, ready to playback as many times as you want.

A couple of hints for creating a good vocal track that will get good results in Wavemotion Analysis:

1. Record in a quiet room.
2. Do not let your recordings clip and get distorted
3. Make sure that you set your levels to record your voice as loud as possible without getting distortion.

That last point is important. If your levels are low, then the background noise in the room is closer to the level of your voice than if your levels are high. You will have to 'normalize' the track when you are done. If the waveform isn't very big (even though you can turn up the volume and hear it fine) it won't give you much jaw movement. You need wavforms that fill the entire bandwidth. If you raise the volume of a track that you recorded at a very low level, the background noise rises with it and gets louder. Aside from sounding bad, this is noise in the track that VSA sees and can't tell that you didn't want it there. That means that the jaw will not close all the way when the speaking or singing stops or in-between words. VSA still sees sound there, even though it is just background noise and uses that sound to program the jaw the same as your voice.

If you have a noisy track, and it is not one that everyone will hear (as is the case with Monster Mash), you can filter out the background noise using a variety of noise reduction techniques depending on the software you are using.

Alternatively, you can manually mute the audio (using your software) in-between every word and sentence to create silent spaces that will allow the jaw to close all the way.

This will make the audio track sound bad and choppy, but if no one is going to hear it, who cares?

This is a great post with a lot of good information. Thanks HB for putting it up.

I like using TrackSkull and the trigger on a joystick to record jaw movement for singing parts where there's music or other sounds in the sound file that VSA interprets as a sound for jaw movement. It takes a little practice but it works good for us.

We are up to 10 talking/singing skulls now and in several of them we were able to set the jaw servos with the same default and max positions so we can use the same jaw motion events for each of those skulls rather than doing each one independently. To avoid the "too synchronized" look and make it look more natural, the motion tracks are copied and pasted and then shifted slightly forward or back on the track to create a slight timing difference. If you want to vary the range of movement between skulls you can also adjust this by changing the % or you could reduce the "maximum" servo position.

For those who use Macs, (we use Macs and PC's) "Garage Band" is a great tool. Garage Band has some pitch tools, timing tools and is really easy to use but it won't export multiple channel sound files (7.1 or 5.1 surround) like Audacity will. We used both for our 2011 sound track. We created a 5.1 channel sound file for 2011 and mounted a speaker in the chest of each talking prop. If Garage Band had the ability to create a 5.1 multi channel sound file I don't think we would have needed Audacity.

Wow Bob what a great thread!!! Thank you for all this work!

Trackskull and using a trigger for the jaw will also work, but it depends on how realistic you want to be.

Trackskull will do the head movements realistically enough if you wear the cap and use the camera, but the joystick method will require more practice. Also, it's a one shot deal. You have to do the whole track at a time, and if you screw up, you start over from the beginning. It can be a big time saver though. It is more difficult to edit the results once it's recorded that way as well if you ever need to.

I am less enthusiastic about using a trigger to work the jaw. You end up without many degrees of motion. Just open or shut. Using the Wavemotion Analysis, it really picks up the subtleties in the voice track. For example, in the following video, watch for how well the jaw tracks the real voice on Sindy. Even catches the vibrato in her voice. Try that with a trigger.
http://www.youtube.com/watch?v=wfzykqCL_n0

In my experience, Wavemotion works much better than TrackSkull when there is only voice present. If there is music, percussion or other sounds, it doesn't work all that well for me. With some practice, I've had good success with the joystick trigger I use. A slight tap of the trigger will open the jaw a small amount where a full pull opens it all the way. It just takes practice to make the jaw appear to match the syllables of the words. Also, where I've had track portions I've wanted to redo in TrackSkull, I've done it setting up TrackSkull to capture the needed motion on an unused VSA track. When the track is imported into VSA, you can cut it from the unused track and paste it onto the track you are editing. Then you just adjust the timing. The vibrato capture is really excellent in the video and you're absolutely right, doing that with a trigger would be next to impossible. If I couldn't do it with Wavemotion because of music or other sound on the audio track, I think I'd just try and do it laying down events manually in VSA. I've never tried it, but I suppose too, I could play such a track over earphones and record only my own voice singing along into a microphone to create a voice only track for Wavemotion Analysis.

Our "Ghost Host" this last Halloween was partly my voice, partly the voice of the Ghost Host from the Haunted Mansion and then he sang Grim Grinning Ghosts with the rest of the skeletons. I used Wavemotion for my voice and the Haunted Mansion voice, the singing part for our Ghost Host I did with TrackSkull. The Haunted Mansion voice had to be edited some in VSA because of the organ in the background and a thunder clap.