Automated transcripts from video with Whisper(.cpp)

I tend to be something of a digital packrat. If there’s interesting data somewhere, I’ll collect it just in case I want to do something with it.

Helpful? Usually not. But it does lead to some interesting scripts.

In this case, I have a site that hosts videos. I want to download those videos and get a text based transcription of them. With new AI tools, that shouldn’t be hard at all. Let’s give it a try!


Genuary 2023.10: Generative Music


Spend a month making one beautiful thing per day, given a bunch of prompts. A month late, but as they say, ’the second best time is now'.

Let’s do it!

10) Generative music

Wikipedia: Generative music

NOTE: Because of limitations in JavaScript autoplaying sound, you must click to start.

NOTE: This currently doesn’t work in Safari. Something funny with the p5.sound addon. I’m working on it.


AoC 2017 Day 18: Duetvm

Source: Duet

Part 1: Create a virtual machine with the following instruction set:

  • snd X plays a sound with a frequency equal to the value of X
  • set X Y sets register X to Y
  • add X Y set register X to X + Y
  • mul X Y sets register X to X * Y
  • mod X Y sets register X to X mod Y
  • rcv X recovers the frequency of the last sound played, if X is not zero
  • jgz X Y jumps with an offset of the value of Y, iff X is greater than zero

In most cases, X and Y can be either an integer value or a register.

What is the value recovered by rcv the first time X is non-zero?


Audiobooks to Podcasts

I’ve recently started to listen to audiobooks again (The Aeronaut’s Windlass). If you buy books through Audible or some other setup that has their own app, it’s a straight forward enough process. On the other hand, if you have them on CD and want to play them on a mobile device… It’s a little more interesting.

I tried a few different apps that purport to do exactly what I wanted: import an audiobook as a folder full of MP3s and play them, but none that quite meet what I wanted. Since I also listen to a lot of podcasts and have more than one podcast app that I really like (I’ve used and liked both Downcast and Pocket Casts), I decided to see if I couldn’t use one of those as an audiobook player.


Making music, part 3: Making noise

Last week we parsed some music. That post was in a bit of a hurry, so we had to leave off a fair few important pieces (like ties and slurs for one; chords for a rather bigger one). We’ll get to them soon, but for now we want to actually get something playing back.


Making music, part 2: Taking shape

It’s been a bit, but as you may have noticed life is a bit mad at the moment. But I’ve still made some progress.

When we left off last time, we’d finished the first step towards making some lovely music with Racket: tokenization. Now we want to take those songs and form them into something actually approaching music.


Making music, part 1: Reading ABC notation

It’s been a bit since I’ve had time to post1, but I’ve got an interesting new project that I’ve been working on. It’s a bit more complicated, ergo spread out over a few posts, but those tend to be the more interesting posts anyway, eh?

The basic idea is that I want to be able to write and play music in Racket. One end goal would be to make a library available for the C211 class to give them something else to work with (in addition to <a href="//"/wombat-ide/c211-image-api/">images and <a href="//"/wombat-ide/c211-turtle-api/">turtles). To that end, here’s my current plan of attack2:

  • Write a lexer for ABC notation to turn raw text into a list of tokens
  • Write a parser to turn those tokens into a song (for example dealing with the interactions between key signature/accidentals and meter/note duration)
  • Use the rsound library on Planet2 / pkg to play back individuals notes and chords
  • Tie it all together to play a parsed song in ABC notation using the rsound library
  • (maybe): Use the rsound library to save ABC files as WAV audio
  • (maybe): Figure out the format and save ABC files as MIDI
  • (maybe): Render songs as music sheets/li>

Sounds like fun! Let’s get started.