@surfmate.team/client-kit

Browser-side computation toolkit — domain-agnostic, zero-dependency. Pure calculations + thin Web Audio actions, layered the Grokking Simplicity way (data / calculation / action). First module: audio.

npm i @surfmate.team/client-kit

audio

import { buildDialogueTracks } from '@surfmate.team/client-kit/audio'

// Two single-person voices → three time-aligned WAV tracks for a
// two-person lip-sync dialogue (left speaks, 1s pause, right speaks).
const { left, right, full } = await buildDialogueTracks(leftBlob, rightBlob, { gapSec: 1 })
//  left  → audio_1 (drives the left person's mouth; speech at t=0, silence after)
//  right → audio_2 (drives the right person; silence until the gap, then speech)
//  full  → the output soundtrack (both voices) + sets total length
// All three are the SAME length and aligned on one timeline.

Lower-level pieces:

import { audioBufferToWav, buildSilenceWav, renderTimeline, timelineDuration } from '@surfmate.team/client-kit/audio'

buildSilenceWav(3)                      // 3s of silent mono 16 kHz WAV (calculation)
audioBufferToWav(audioBuffer)           // AudioBuffer → 16-bit PCM WAV Blob (calculation)
renderTimeline(placements, total, sr)   // place buffers on one track, render (action)

Composes directly with @surfmate.team/digital-human-runninghub's createRunningHubDialogueVideo — buildDialogueTracks produces exactly its leftAudio / rightAudio / fullAudio inputs.