Lotus CIandClaude Opus 5.5
1ebc42863e
CI / Build & Quality Checks (push) Skipped
CI / Secret scan (gitleaks) (push) Skipped
CI / Playwright smoke (e2e) (push) Skipped
CI / Build & Quality Checks (pull_request) Successful in 1m40s
CI / Trigger Desktop Build (pull_request) Skipped
CI / Docker image build & smoke test (pull_request) Skipped
CI / Secret scan (gitleaks) (pull_request) Successful in 6s
CI / Playwright smoke (e2e) (pull_request) Successful in 19m45s
The CPU column was guessed (DeepFilterNet "25-50%", DTLN "10-20%") with no unit, so it read like Task Manager's whole-CPU figure and looked wildly off. Measured with each model running exactly as Lotus Call runs it (real-time speech, browser process CPU time from /proc minus a no-model baseline, Ryzen 7 PRO 4750GE): DeepFilterNet ~30%, DTLN ~9%, RNNoise ~3%, Speex ~2% of one core. Label it "one core" and explain the Task Manager relation. DTLN's download was listed as ~4 MB but fetches ~20 MB (17 MB worklet with the runtime inlined, plus 4 MB of models); DFN's ~18 MB is right (17.6 MB). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_017PCb59uY3UKqDqt7B6hzd5
102 lines
3.6 KiB
TypeScript
102 lines
3.6 KiB
TypeScript
/**
|
|
* Detection utilities + model catalog for Lotus ML noise suppression
|
|
* (DeepFilterNet 3 / DTLN / RNNoise / Speex). The catalog is ordered by
|
|
* quality (and, correspondingly, CPU cost) — highest first — and drives the
|
|
* order of the model dropdown in settings.
|
|
*/
|
|
|
|
import { DenoiseModelId } from '../state/settings';
|
|
|
|
// CPU = share of ONE core while in a call, measured with each model running
|
|
// exactly as Lotus Call runs it (Ryzen 7 PRO 4750GE, Chromium); download = bytes
|
|
// actually fetched on first use. Re-measure when a model or its settings change.
|
|
export type DenoiseModel = {
|
|
id: DenoiseModelId;
|
|
name: string;
|
|
description: string;
|
|
cpuUsage: string;
|
|
binarySize: string;
|
|
transients: 'Poor' | 'Good' | 'Excellent';
|
|
voiceQuality: 'Moderate' | 'High' | 'Very High';
|
|
};
|
|
|
|
// Ordered best-quality (highest CPU) first — this is the dropdown order.
|
|
export const DENOISE_MODELS: DenoiseModel[] = [
|
|
{
|
|
id: 'deepfilternet',
|
|
name: 'DeepFilterNet 3 (beta)',
|
|
description:
|
|
'Studio-grade deep-learning model (48 kHz fullband, ONNX). Best quality; highest CPU and a larger one-time download.',
|
|
cpuUsage: '~30%',
|
|
binarySize: '~18 MB',
|
|
transients: 'Excellent',
|
|
voiceQuality: 'Very High',
|
|
},
|
|
{
|
|
id: 'dtln',
|
|
name: 'DTLN (beta)',
|
|
description:
|
|
'Dual-signal deep-learning model (16 kHz). Strong on transient noise; moderate CPU.',
|
|
cpuUsage: '~9%',
|
|
binarySize: '~20 MB',
|
|
transients: 'Excellent',
|
|
voiceQuality: 'High',
|
|
},
|
|
{
|
|
id: 'rnnoise',
|
|
name: 'RNNoise',
|
|
description:
|
|
'Lightweight hybrid model (48 kHz). Very low CPU; good for steady noise like fans, but can sound processed at full strength.',
|
|
cpuUsage: '~3%',
|
|
binarySize: '< 1 MB',
|
|
transients: 'Good',
|
|
voiceQuality: 'Moderate',
|
|
},
|
|
{
|
|
id: 'speex',
|
|
name: 'Speex (Legacy)',
|
|
description:
|
|
'Classic DSP noise suppressor. Minimal CPU, gentlest on voice; weakest suppression.',
|
|
cpuUsage: '~2%',
|
|
binarySize: '< 1 MB',
|
|
transients: 'Poor',
|
|
voiceQuality: 'Moderate',
|
|
},
|
|
];
|
|
|
|
export const isMLDenoiseSupported = (): boolean => {
|
|
if (typeof window === 'undefined') return false;
|
|
|
|
// Requirements:
|
|
// 1. AudioContext/webkitAudioContext (Web Audio API)
|
|
// 2. AudioWorklet (Real-time processing in a background thread)
|
|
// 3. getUserMedia (Microphone access)
|
|
const hasAudioContext = !!(
|
|
window.AudioContext || (window as Window & { webkitAudioContext?: unknown }).webkitAudioContext
|
|
);
|
|
// Use `typeof` rather than `!!AudioWorkletNode`: on a browser that has
|
|
// AudioContext but no AudioWorkletNode binding (older Safari/Edge), a bare
|
|
// reference throws ReferenceError, making this feature-detection helper throw
|
|
// instead of returning false.
|
|
const hasAudioWorklet = hasAudioContext && typeof AudioWorkletNode !== 'undefined';
|
|
const hasGetUserMedia = !!(navigator.mediaDevices && navigator.mediaDevices.getUserMedia);
|
|
// Every ML model compiles WebAssembly (and DFN/DTLN load worklets via blob
|
|
// URLs). Under a strict CSP without `wasm-unsafe-eval` (e.g. some desktop/Tauri
|
|
// shells) WASM is unavailable, so gate on it — otherwise we'd offer ML and then
|
|
// silently fall back to the raw mic in-call.
|
|
const hasWasm =
|
|
typeof WebAssembly !== 'undefined' && typeof WebAssembly.instantiate === 'function';
|
|
|
|
return hasAudioWorklet && hasGetUserMedia && hasWasm;
|
|
};
|
|
|
|
/**
|
|
* EXACT requirements for ML Denoise (for UI display).
|
|
*/
|
|
export const ML_DENOISE_REQUIREMENTS = [
|
|
'Modern browser with Web Audio API support',
|
|
'AudioWorklet support (Chrome 66+, Firefox 76+, Safari 14.1+)',
|
|
'WebAssembly (WASM) support',
|
|
'Microphone access',
|
|
];
|