Templates, sectioning, custom workflow for music video creators?

[xanmankey]

First of all, I want to say that this is an incredible library! As someone interested in making music videos, there's a lot of potential here to automate the process of editing and putting together such videos. With that in mind, I would like to contribute extra support in the form of the following:

• moviepy.video.VideoClip: a substituteAudio field for the VideoClip class, a sync_audio_and_video function
• moviepy.audio.tools: a sync_audio function
• moviepy.audio.fx: a high_pass function, a low_pass function
• moviepy.video.section: a class for sections (a grouped collection of VideoClip objects), a determine_music_sections function
• moviepy.video.template: a class for templates (an automated way of editing sections)

Here's an example process of how I would use these features:

• Create my audio and video clips from the recorded files in the respective directories (I record two different audio tracks, one associated with the video that is lower quality and one using an external microphone). I would make sure to specify the audio clip I want for my track in the newly created substituteAudio field
• Because the substituteAudio field is not None, the sync_audio_and_video function is called using the current object, which would use librosa's support for Dynamic Time Warping (I still need to do more research for this) to find the interval where the production audio clip best matches the video audio clip, calls cutout to only retrieve that section of the video (thereby automatically removing extra takes that would otherwise need to be manually edited out) and replaces the audio
• Then, once I have my cropped clips with production audio I need to go about creating my video. That's where determine_music_sections would come in. In the context of music videos, it would take into account beat, energy, and harmonic metrics to determine when the song transitions to a new idea. Depending on the type of videos, a video could be sectioned differently, although I'm not yet sure how to scale or adapt this implementation. But the general idea is that sections have some kind of association to a Template class, which contains a bunch of default Templates that can be applied for the specific section of the video.

Templates are something I personally view as very important to moviepy, specifically because there's limited support for templates in video editing as a whole (or at least I've personally found that to be the case). Both the Templates and Sections classes would need to be very flexible and open to contribution to handle different kinds of videos, and yet flexible and simple enough to streamline the process. It's a tough balance and I'm not sure how I would implement it.

• Finally, the high_pass and low_pass effects I just think would be nice to have for video editing; in some cases, applying high_pass and low_pass filters is better suited to the video editing process than the actual music production and post production process

Implementing these features would introduce librosa as a dependency for the project, which I think is acceptable given the utility that librosa can provide for audio editing and analysis. I just wanted to put my ideas out there, I started implementing some of them in the process of creating a library of my own, but when I saw that moviepy was in need of contributors I wanted to at least mention them here; I'm not sure if they align with the current goals of the project or where the project is currently at, but I would love to at least give implementing these a try.

Also I typed this out fairly quickly and I just recently had these ideas, so feel free to ask for clarification or bring up potential issues with this approach