Available Models for HARP

The following models are always available for use within HARP, and can be found in the drop-down menu at the top of the main panel:

• Stable Audio (text-to-audio) (stability/text-to-audio) can generate music, sound effects and/or soundscapes, based on a text description.

• Stable Audio (audio-to-audio) (stability/audio-to-audio) enables the transfer of style or creation of variations based on text and audio conditioning.

• Text2Midi (teamup-tech/text2midi-symbolic-music-generation) generates MIDI files from textual descriptions.

• Demucs (teamup-tech/demucs-source-separation) performs source separation on music, splitting it into "Drums", "Bass", "Vocals", and "Instrumental" stems.

• High Resolution Piano Transcription (teamup-tech/solo-piano-audio-to-midi-transcription) converts audio of solo piano playing into a corresponding MIDI file.

• Anticipatory Music Transformer (teamup-tech/anticipatory-music-transformer) performs harmonization on MIDI inputs, generating additional notes to provide the harmony for a given melody.

• VampNet (teamup-tech/vampnet-conditional-music-generation) generates variations or "vamps" on music audio, and offers a variety of controls for determining how the vamps diverge from the original input audio.

• Harmonic/Percussive Source Separation (teamup-tech/harmonic-percussive-separation) performs source separation on music by splitting it into "harmonic" and "percussive" tracks, allowing for the extraction of drum-like elements.

• Text-to-Speech (teamup-tech/Kokoro-TTS) generates speech in the style of a chosen voice preset given a text prompt.

• Voice Cloning (teamup-tech/MegaTTS3-Voice-Cloning) generates speech conditioned on another speech recording following a text prompt.

• Midi Synthesis (teamup-tech/midi-synthesizer) synthesizes MIDI into audio using the standard MuseScore SoundFont.