Midwesing

Conversation in,
Midwest emo out.

Upload a recording of people talking. Midwesing listens to the rhythm, the pauses and the rise and fall of the voices, then writes a Midwest emo guitar part underneath — arpeggiated, a little twinkly, always quieter than the conversation itself.

Everything runs locally in your browser. Your file is never uploaded anywhere, and the original conversation stays the primary audio in the final mix.

1 · Upload

Drop a file here, or click to choose

Audio: WAV, MP3, FLAC, M4A, OGG · Video: MP4, MKV, MOV, WEBM

2 · Arrangement preset

Same file + same seed + same preset = same arrangement.

Speech stays primary. Changing this after generating re-mixes instantly.

Midwest emo lives in the room. Default is high on purpose; re-runs the chain after generating.

Dotted-eighth delay at the preset tempo, with audible repeats — not a hint of delay.

3 · Generate

    4 · Preview & download

    Nothing here yet. The processed preview and the WAV download appear after a successful generation.

    How it hears you

    Vocal pitch movement→Melody contour, snapped to scale
    Speech activity→Note activity
    Pauses→Rests and sustained notes
    Speech intensity→Note velocity
    Speech rhythm→Note duration and placement

    Pipeline: validate → extract audio → voice activity → pitch & rhythm → musical mapping → melody → chords → guitar pattern → electric render → effects chain → mix → WAV. If pitch tracking fails, generation falls back to timing and rhythm alone.

    The tone — electric, not acoustic

    No synth pluck and no acoustic soundfont. The source is a real sampled clean electric guitar — actual recorded notes, pitch-shifted at most two semitones — fingerpicked in character, bright and slightly compressed, in the direction of American Football and TTNG. The default preset even tunes to Open F (F A C F A C), so open strings ring through the chords the way they do on the records. Then the full chain runs, with reverb and echo well up, before it ever meets the conversation:

    Electric guitar (sampled)→Fingerpicked arpeggios, open strings, tapping
    Compressor→Evens out the picking, slight squeeze
    EQ→Low cut, presence lift, bright top
    Subtle overdrive→Low gain, soft clip, mostly clean
    Chorus→Slow, shallow — width, not warble
    Echo (delay)→Dotted-eighth repeats you can actually hear, adjustable
    Reverb→A big room around the notes, adjustable, then into the final mix

    Stereo: the melody layer sits slightly left, the arpeggio layer slightly right. Distortion stays low-gain and optional in spirit — the drive here never gets past a soft edge.