Extensions:2.6/Py/Scripts/Import-Export/Import Movement Fom Audio File

Import Movement Fom Audio File
Just Select a Object, go to Object tab - Click> Import Movement From Audio File - Click> Import Audio File - and finally Click> - Import Key Frames

UI location	Inside Object Tab
Usage	Just Select a Object, go to tab - Import Movement From Audio File - Click in - Import Audio File - and then - Import Key Frames.
Version	0.21	Author(s)	Vlassius
Blender	2.57	License	GPL
Category	Import-Export	Distribution	Extern

Executable information

File name	import_sound_to_anim.py
Current version download	(not yet)
Python modules	wave
External Python Modules or dependencies	none
Data	*.wav

Warning	Depending of your computer power and the size of the .wav file, the process can take a long time.
Links	http://wiki.blender.org/index.php/User:Vlassius
Release Log	Here
Known Issues	(please, tell me)

Import the movement of a .wav file to a animation - Engine

Hi,

Maybe this can help some peoples that needs to deal with sounds and needs to sync it to a movement. This script takes a .wav file and transforms it in movement into Blender (key frames) using the amplitude of the audio.

Examples of use:

• Make balls jumping with the sound.

• Make a Sound System

• Make a business training to learn difficult concepts.

• Make a hand move and fake a piano player.

• You have some minutes of audio to make your robo "talk". If you have the audio file (just with voice), you can get a very good sync to the mouth (using bones).

I used it a lot of times and got good (and funny) results mainly to Business Training where is just necessary a basic sync and not a letter-by-letter sync.

The result is very quick and simple comparing with other methods.

Get Started Manual

1- Select a object, you can use the default cube (Mesh, Bone, Empty or Camera) 2- Go to the Object Tab 3- Go to "Import movement From a Wav File" 4- Go to "Direct to a property"

Let everything "as is" for now.

Click the button "Process Wav" and open a file. Must be a 16 bit wave file (you can use a program like Audacity to save in this format.

(It works if there is just one voice or one instrument at a time. The script takes the sound of the file but, it cannot separate for example voice and guitar) See the terminal window about the process, wait until it ends. In a 100KB file, it will take some seconds. In a 20MB file, it will take some minutes.

After this, a button and more options will appear. Let everything "as is" and click "Import Key Frame"

At the end of the process, you will have a text between the two buttons indicating the number of frames imported.

Your "Time Bar" will have the Key Frames.

Try to use a file with some seconds to see the performance. It can take several minutes to process and import a large file. I've processed a 80MB file (7 minutes of audio) into a MESH to change its location in just 3 minutes but, the same file in a bone to deform a mesh took 15 minutes. The major time is spend in import the values to Blender.

Time to have a fun, press Control A to start the animation and see your object moving.

Basic Options:

General Information

1- Audio Sensitivity 1 to 6. See this in detail at next section.

2- Frames/s: Setup of your scene. Necessary to correct sync the frames. Automatically imported. Bad values will make loose of sync. Just let it the same that you use in your scene. Advanced- Can be useful to change it if you use multiple of the original values.

3- Acts/s: Max frames per second to put the key frames, 1(one) will be one key frame inserted per second; 10 will be ten key frames inserted by second. For example, to a mouth you probably want 4 and, to a arm ore hand, probably you will use 1 because mouth movements are quick and arm/hand movements are generally soft and slow. Min:1 Max:Num. Key Frames/s - Default:4

4- Scale: If you want to scale the values of samples you get. If 1: gets 0 ~255. If 100. gets 0~2.55. If 1000, gets 0~0.255. To a mouth bone, maybe you will use a Scale of 2000 and 100 to test a default cube.

• If your animation has a ball to fly 25 units of Blender, set this to 10
• If your animation has a ball to fly 12.5 units of Blender, set this to 20
• If your animation has a ball to fly 2.5 units of Blender, set this to 100
• If your animation has a ball to fly 1.25 units of Blender, set this to 200
• If your animation has a ball to fly 0.25 units of Blender, set this to 1000
• If your animation has a ball to fly 0.12 units of Blender, set this to 2000

5- Apply the movement:

• Change Location:
  • Property: Location - and choose one or more location axis.

• Change Scale:
  • Property: Scale - and choose one or more location axis.

• Change Rotation:
  • Property: Rotation - and choose one or more location axis.

6- Initial frames to put the key frames.

7- Clip Min: Cut the lower value to the sample. Samples goes from 0 but if set here 100, any value less than 100, will be set to 100.

8- Optimization: That is like a round to the samples values. If bigger, more quick the import but less precision. If 0, hi precision values and more slow to process. Default 10. Min:0 to Max:255.

9- Clip Max: Max value for the samples. Default: 255. If this number is lower than 255, any sample value higher than this value will be set to it.

A- Hard Transition: If you want movements like a month (in movement just when speaking) check this. If you want a soft transition to the movements, like a arm moving, leave in blank. Default: checked.

B- D- C- Offset to X, Y and Z. This value will be added to the axis with the result value of sound file.

E- Button to open and process the wave file.

Detailed: Adjust Sensitivity

The sound file may contain a very loud (high amplitude) sound, or a very low (low amplitude) sound or any in between. To adjust this, you will need to adjust the sensitivity to get most sample amplitudes fits inside the samples (0 to 255).

• If you use a very low amplitude sound with a very low sensitivity, you you get the samples very with low amplitude like 0 to 50 instead of 0 to 255. If the samples is very low like this example, you will loose movement because the scale is squeezed from 0 to 50.
• If you have a very loud sound and you set the sensitivity to max (6), you will get the samples squeezed again to the high part of scale, for example 200 to 255. In this case, you will loose movement too.
• The right way to do it in most times is to put the sound samples fit inside 0 to 255 values.

The best way to do it is using a tool that meter the result, and see if it is distributed inside the 0 to 255 values. It seems very technical matter but, the script comes with a graph meter to help you to do it very easily.

All you need to do is start Blender in a terminal window of your OS. http://wiki.blender.org/index.php/Doc:2.6/Manual/Interface/Window_system/Console_window

Use your Blender normally with the terminal window open or minimized. When the time to use the script comes, before open the audio file, open the terminal window thus you can see what happens.

OK, lets see how to do it in a very simple way:

• Just checking, you started Blender using a terminal window. Choose your object to receive the Key Frames, select it, go to "Object" tab, at the end of the tab open the "Import Movement From the Wav File", click the button "Direct to a Property".

• As we want to test just the sensitivity, please, leave all settings "as is".

• Click the button "Process Wav", open your 16 bits sound .wav file and look at the terminal window. The next picture is a example of what you will see.

• Look at the "Graph Value". In this example, you can see that the levels (marked with *) go from 0 to something like 60% of the bar. This mean that the level is almost good, but can be better. Just to clarify, 60% of the bar means a level of 60% of 255, that is about 150, the samples are going from 0 to 150 and we want it 0 to 255.

• OK, so now we will increase the sensibility to the levels go up. I just changed the "Audio Sens" to 2. Lets see what happened.

• Very good. The level of samples goes from 0 to almost 255 (almost 100% of the bar graph). But what happens if i set the "Audio Sens" to 3? Lets see it.

• Now, the samples goes from 0 to 255 but, you can see that almost all samples are in 100% (255). This will make your movement goes from 0 to 255 without values between.

In most cases, this is a problem but, you may want just use exact like this like a ball that goes up on a beat and goes down when there is no sound.

• OK then, you can choose the sensitivity to meet your needs. Now just to complete the example, the next two images are examples to the import process when you click the next button to import to Blender.

Examples

Ball

I want a ball that stay stopped until a beat and goes to the maximum value and goes back, I mean, I want just the ball down until the beat happen. The basic setup to this is: Everything default but:

• Disable "Hard Transition" (or maybe not, see what you get in both cases)
• Adjust the audio sensitivity to get values from 0 to 255 without values in between. To do this, see the section "Adjust Sensitivity" here.

Advanced:

• You can set the "Clip Max" to 100 for example to have sure that no values in between would bottler you.
• Set the scale to 1 to have movements from 0 to "Clip Max". Just adjust it as you need.

Hand in Piano

I have a scene that shows the hand and a piano but does not show the fingers, can I sync the play with the movement of the hand?

• Sure, just take your audio with just the piano sound, apply to the hands.
• Adjust the audio sensitivity.
• Adjust the "Scale", probably something like 500 or 1000 to make little movements.
• You can use with or without "Hard Transition". If Checked the movements will be appear more stronger, with quick movements. If not checked, the movements will be softer.

Finger Playing Piano

I want to a finger touch a key when a sound plays.

• Its very simple but, maybe you need some ground here.
• You need to remember that the script will not isolate the sound that you like. You need a isolated sound (the sound that you want to move the finger) and apply to the finger.
• Put a bone in your finger.
• You will apply the script to the bone.
• Adjust the "Audio Sens" to have a good range of 0 to 255 but without values in between as the finger is stopped or touching the key, nothing in between. You can use the "Clip Max" to 100 for example to help you to this step.
• Change the "Property" to "Rotation" and choose the proper axis. The help discover the proper axis, change the rotation to your bone manually and see where to apply the values.
• Once you get the right setup for one finger, you can use the same setup to the others fingers.
• You can apply the values of all fingers to the hand thus, the hand will move too when the finger moves.
• As a additional step, you can use a program to filter the audio and make the setup a lot easier. Just use a high filter in this case and use the resultant audio to use the scripts. This will make the piano key sound like a "tick". You can use Audacity to easily do it. That was what i done in the example video.

About the example of a piano player (start of this page): The script was used 13 times: 10 fingers, 2 hands and the keyboard No adjust was done after import the movements. All movements was imported and processed in only 2 hours. There are more than 9000 keys (movements) inserted into the animation. The setup took no more than one hour. That was almost impossible to do it manually and really impossible to do it in one day without the script.

Tutorial - Bubble Man

In this example you will have all steps and files to make this animation:

Files

ZIP Sounds File - 18MB

Blender File - 730KB

How To Do It

OK, I said all steps, lets go...

Make a character, in this case, Bubble Man. (half hour)

• Prepare to receive the movement, in this case i use a Meta Ball to the head and this script does not work with it. Problem? Not at all.
• To apply the movement to the Meta Ball you can place a cube at zero "z" and out of camera view, then made parent the cube and the Meta Ball so you can apply the movements to the cube and the Meta Ball will follow.
• The cube is at zero "z" because the script apply absolute values, not relative to the position of the object, I mean, if the head is at 2.00 "z", after apply the script it will be from 0 to max value, where max value is the sample value times scale used (and can be negative, once you can apply to -z) and using a external parented object, is possible to make the setup easier.
• Now we have the target object (cube) to receive the movements.

Make a Sound to use. You can record voices or instruments. In this case it was used a piano keyboard to generate the sound. ( half hour)

• Some simple sound recorded in the keyboard and saved as MIDI file.
• Using a program called Rosegarden, I made a other MIDI file (using the first MIDI) just with one instrument. MIDI file is a data file that has instrument, notes, etc... You can isolate one or more instrument, edit then, etc...
• Again in the piano keyboard, the MIDI files was exported to .wav files. One MIDI exported the entire music to the animation, the second MIDI was just the choose instrument to generate movement.

Prepare the sound to generate the movement. (ten minutes)

• We have now a .wav file with just one instrument (acoustic snare) and the sound is not very clean to generate the movement as I want a "beat or not beat".
• To filter this .wav it was used the program Audacity and applied a "High Band Filter" at 10Khz.
• The final sound is very simple and clean to make the desired result.

Apply the Movement to the target. (five minutes)

• Open Blender (if possible, open in a terminal window), Select the cube (parented with the head).
• Go to script panel,
  • change sensibility to 2
  • Location to -z
  • Scale 130
• Click: Process Wave. Take a look in the terminal window.
• Click: "Import Key frames". Take a look in the terminal window.
• Adjust the "End Frame" if necessary (scene tab).
• Ready to render!!!

Total time (without render): less than 1.5 hours

Details: After the frame 1300, I made simple movements to end the animation to fit the music. The render time in my computer is 40 sec per frame, it takes about 14 hour to render. You need a Blender with Cycles enabled to render this.

Final Message

Use your imagination, have fun!!!!!!!!!!!!!!!

.