Yesterday, I shared my first forays into Vibe Coding with Anthropic’s Claude. Vibe coding is the term for using AI to create small web apps for a specific task. While some coding skills are helpful, they are not necessary. It’s more that you need to refine your instructions and revise the AI’s work until you get to something close to what you want to have created.
With Claude, it builds the apps right in it window and then provides a preview, and then hosts the app online, when you hit the publish button. I am not sure if other AI apps do it the same way or not.
I decided to try to get a little deeper into vibe coding, using a series of recent morning Haiku poems as the text. I did a few experiments to work with the poems (such as randomizing phrases to create new poems, and switching lines of the poems with other lines of poems to create new poems) but then a side experiment with an interesting response by Claude sparked an idea around the concept of word-to-sound.
I went into that word/sound/visualization rabbit hole, trying to get at a way to create an app that connected my poems to sound visualization.
It worked, and I think it’s pretty nifty.
Here are the notes from Claude on what it was doing as it was coding:
I’ve created an expanded application that explores the connection between poetry and sound through four different audio-inspired visualizations:
- Enhanced Waveform Visualization
- Creates dynamic sound waves based on syllable patterns and sound types
- Analyzes hard consonants (k, t, p, d, g, b) vs. soft consonants to shape the wave amplitude
- Displays both the wave pattern and a filled area representing sound intensity
- Shows vowel/consonant ratio and hard/soft sound distribution for each poem
- Musical Staff Visualization
- Translates each poem into musical notation on a standard five-line staff
- Converts words to musical notes with:
- Note position based on vowel concentration
- Note duration (whole, half, quarter, eighth) based on syllable count
- Accents based on presence of hard consonants
- Preserves the lyrical structure by showing words under their corresponding notes
- Frequency Spectrum Visualization
- Analyzes the distribution of letters as if they were sound frequencies
- Displays vowels and consonants with different intensities of the same color
- Shows the “sonic texture” of each poem through letter frequency analysis
- Includes metrics on vowel/consonant balance and hard/soft sound distribution
- Phonetic Rhythm Visualization
- Maps the natural rhythmic patterns in each poem
- Represents strong (stressed) and weak syllables as different-sized nodes
- Shows how words connect rhythmically along a timeline
- Groups words into rhythm patterns to reveal the poem’s natural cadence
Each visualization uses sound-related principles to transform your poems into visual representations that capture their acoustic qualities. The app analyzes:
- Syllable stress patterns (strong vs. weak)
- Vowel and consonant distribution
- Hard vs. soft consonant sounds (which affect perceived texture)
- Natural rhythmic groupings
These visualizations help reveal the musicality inherent in your poetry, demonstrating how the sound patterns contribute to each poem’s unique character and flow.
Peace (sounding like poems),
Kevin