MarkoVid

MarkoVid running on a Prof. Paoletti EECS 281 Lecture
The Python terminal running MarkoVid, with sentence history

Inspiration

We thought Markov Chains were really interesting and thought it would be funny to combine these with video transcription.

What it does

Option to take in a word or phrase and generate this segment of video or use a Markov Chain Generator to generate several sentences that contain words from the video.

How we built it

We used Python to edit videos and put the video into clips and the Google API to find where words appeared in the video.

Challenges we ran into

Getting the video and audio into the right formats and having words not either cut off or have a pause after them.

Accomplishments that we're proud of

We put some online videos and lecture videos from our professors through it and had some hilarious results.

What we learned

We learned about Python, the Google Cloud API, Markov Chains, text-to-speech, and automated video splicing.

What's next for MarkoVid

Add an automated process to download videos from the Internet and put them in the right format with correct sampling rate and mono/stereo audio.

Built With

Submitted to

MHacks 11

Created by

I came up with the idea, and I used the Markovify library for Python to create sentences from video transcripts generated by the Google Cloud API. From there, I used the word timestamps it also generated to piece together audio/video clips from the original source videos that match the words and sentences generated by Markovify. I put captions on the videos to make them more legible (since the audio clips are often hard to comprehend).

Ian Robinson
I worked on the front end and creating an application using React Native that could connect to a back-end server and Google Cloud storage. The application worked with both native video input and text/url input. I never worked with React Native before but I leaned a lot about cross platform development, javascript, and APIs.

Junlin Mo
Helped research video splitting and made manual user word/phrase input option.

Helena Gregg
I worked on the Google Cloud integration and using the Google voice transcription API to create transcripts of videos to feed the Markov chain generator. I also worked on allowing the user to enter their own phrase and have the video spliced accordingly.

Nicholas Kroetsch

Updates

Junlin Mo started this project — Oct 14, 2018 06:36 AM EDT

Leave feedback in the comments!

Log in or sign up for Devpost to join the conversation.