DemonGatanjieuandClaude Opus 5.5
0c1287ddc2
Voice-over's side panel shows who reads, speed, the language (one line while detected) and the tone with a listen button; the comparison, the numbers and the links fold under More. GPT-SoVITS Settings opens with one status line and the tone to use; the language rules, base files, places and version are folded and open by themselves when they need the user. Plainer words throughout: tone, Steady / Natural / Lively, As the character, Always read like this, Own settings, base files. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
96 lines
4.3 KiB
JavaScript
96 lines
4.3 KiB
JavaScript
import { t } from './interface_settings.js';
|
|
import { anomalousAlert } from './ui_dialog.js';
|
|
import { button, el } from './usage_view.js';
|
|
import { buildTtsPrompt } from './audio_script.js';
|
|
import { startPromptJob } from './audio_tts_run.js';
|
|
|
|
/**
|
|
* Listening to the generation modes: one sentence read by one character with the same seed,
|
|
* only the mode changing, through the node's Preview Audio (temp files, out of the audio
|
|
* gallery). Takes are kept for the page session by character, sentence, language, speed and the
|
|
* mode's numbers, so a mode heard before plays again at once and a changed number makes a new
|
|
* take; one take is made at a time.
|
|
*/
|
|
|
|
const SEED = 20261017;
|
|
|
|
/** `onChange()` when a take starts or is done (the caller draws its row again). */
|
|
export function createModePreview(onChange) {
|
|
const takes = new Map(); // key -> url
|
|
const player = new Audio();
|
|
let job = null;
|
|
let making = null; // the key of the take being made
|
|
|
|
const keyOf = take => JSON.stringify([take.character, take.text, take.language, take.speed, take.mode, take.numbers]);
|
|
|
|
/** `take`: `{character, text, language, speed, mode, preset, numbers}`; `preset` empty = Auto. */
|
|
async function play(take) {
|
|
const key = keyOf(take);
|
|
player.pause();
|
|
if (takes.has(key)) {
|
|
player.src = takes.get(key);
|
|
player.play().catch(() => {});
|
|
return;
|
|
}
|
|
if (job || !take.text || !take.character) return;
|
|
making = key;
|
|
onChange?.();
|
|
const chosen = take.preset ? { preset: take.preset } : {};
|
|
job = startPromptJob(buildTtsPrompt({
|
|
character: take.character, speech: take.text, seed: SEED, language: take.language, speed: take.speed, preview: true, ...chosen,
|
|
}));
|
|
try {
|
|
const { outputs } = await job.result;
|
|
const audio = outputs['2']?.audio?.[0];
|
|
if (!audio) throw new Error(t('scriptRunNoAudio'));
|
|
const query = new URLSearchParams({ filename: audio.filename, subfolder: audio.subfolder || '', type: audio.type || 'temp' });
|
|
takes.set(key, `/view?${query}&t=${Date.now()}`);
|
|
player.src = takes.get(key);
|
|
player.play().catch(() => {});
|
|
} catch (error) {
|
|
if (!error.cancelled) await anomalousAlert(t('scriptRunFailed', { error: error.message }));
|
|
} finally {
|
|
job = null;
|
|
making = null;
|
|
onChange?.();
|
|
}
|
|
}
|
|
|
|
/**
|
|
* The listen row: "Listen to this one" for the picked mode, then the modes already heard
|
|
* with this sentence (a click picks the mode and plays its take). `takeFor(id)` builds a take;
|
|
* `lead`: elements before the button (which line is read); `withButton` false leaves the
|
|
* button to the caller (the Script Director's cards) and says "Making it…" instead; `label`:
|
|
* the button's words before a take exists.
|
|
*/
|
|
function row({ takeFor, selected, modes, onPick, disabled = false, lead = [], withButton = true, label: first = 'ttsListen' }) {
|
|
const box = el('div', 'anomalous-tts-takes');
|
|
box.append(...lead);
|
|
const current = takeFor(selected);
|
|
const key = keyOf(current);
|
|
if (withButton) {
|
|
const label = making === key ? 'ttsListenMaking' : (takes.has(key) ? 'ttsListenAgain' : first);
|
|
const listen = button('anomalous-sd-btn', t(label), () => play(current));
|
|
listen.title = t('ttsListenHint');
|
|
listen.disabled = disabled || (Boolean(job) && !takes.has(key)) || !current.text || !current.character;
|
|
box.append(listen);
|
|
} else if (making === key) {
|
|
box.append(el('span', 'anomalous-tts-setup-meta', t('ttsListenMaking')));
|
|
}
|
|
const heard = modes.filter(mode => takes.has(keyOf(takeFor(mode.id))));
|
|
if (heard.length) {
|
|
box.append(el('span', 'anomalous-tts-setup-meta', t('ttsListenHeard')));
|
|
for (const mode of heard) {
|
|
const take = button(`anomalous-tts-listen-take${mode.id === selected ? ' is-current' : ''}`, `▶ ${mode.label}`, () => {
|
|
if (mode.id !== selected) onPick(mode.id);
|
|
play(takeFor(mode.id));
|
|
});
|
|
box.append(take);
|
|
}
|
|
}
|
|
return box;
|
|
}
|
|
|
|
return { row, play, stop: () => player.pause() };
|
|
}
|