628 lines
35 KiB
TypeScript
628 lines
35 KiB
TypeScript
"use client";
|
||
|
||
import { useEffect, useMemo, useRef, useState } from "react";
|
||
import { ArrowRight, WaveIcon } from "@/components/icons";
|
||
import { BrandIcon } from "@/components/brand-icon";
|
||
import { SectionHeading, StatusCard } from "@/components/ui";
|
||
|
||
const ttsBase = (process.env.NEXT_PUBLIC_TTS_BASE_URL ?? "/tts").replace(/\/+$/, "");
|
||
const pausePattern = /<pause:(\d+)>/g;
|
||
const typoRules = [
|
||
{ wrong: "爷门儿", right: "爷们儿" },
|
||
{ wrong: "在见", right: "再见" },
|
||
{ wrong: "以经", right: "已经" },
|
||
];
|
||
|
||
type Voice = {
|
||
id: string;
|
||
name: string;
|
||
desc?: string;
|
||
category?: string;
|
||
gender?: string;
|
||
genre?: string;
|
||
accent?: string;
|
||
style?: string;
|
||
styles?: string[];
|
||
preview?: string;
|
||
};
|
||
|
||
type VoiceResponse = { categories?: string[]; facets?: Record<string, string[]>; voices?: Voice[] };
|
||
type TtsConfig = {
|
||
model?: string;
|
||
formats?: string[];
|
||
default_format?: string;
|
||
speed?: { min?: number; max?: number; default?: number };
|
||
pause?: { supported?: boolean; min_ms?: number; max_ms?: number; presets?: { ms: number; label: string }[] };
|
||
};
|
||
type Result = { url: string; fileName: string; format: string };
|
||
type SelectedPause = { start: number; ms: number };
|
||
|
||
const pauseOptions = [500, 1000, 2000];
|
||
const symbolPauses: Record<string, number> = { "，": 300, "。": 800, "！": 800, "？": 800 };
|
||
const erhuaOptions = ["关闭", "弱", "标准", "强"];
|
||
|
||
function normalizeText(value: string) {
|
||
return value.normalize("NFC").replace(/\r\n?/g, "\n");
|
||
}
|
||
|
||
function apiUrl(path: string) {
|
||
return `${ttsBase}${path}`;
|
||
}
|
||
|
||
function escapeHtml(value: string) {
|
||
return value.replace(/&/g, "&").replace(/</g, "<").replace(/>/g, ">").replace(/"/g, """);
|
||
}
|
||
|
||
function renderTextHtml(value: string) {
|
||
let html = escapeHtml(value);
|
||
for (const rule of typoRules) {
|
||
html = html.replaceAll(escapeHtml(rule.wrong), `<span class="typo-mark" data-typo-right="${rule.right}">${escapeHtml(rule.wrong)}</span>`);
|
||
}
|
||
return html.replace(/\n/g, "<br />");
|
||
}
|
||
|
||
function renderEditorHtml(value: string) {
|
||
let html = "";
|
||
let cursor = 0;
|
||
const matcher = new RegExp(pausePattern.source, "g");
|
||
let match = matcher.exec(value);
|
||
while (match) {
|
||
html += renderTextHtml(value.slice(cursor, match.index));
|
||
html += `<span class="pause-token" contenteditable="false" data-pause-ms="${match[1]}" role="button" tabindex="0">⏸ ${Number(match[1]) / 1000}s</span>`;
|
||
cursor = match.index + match[0].length;
|
||
match = matcher.exec(value);
|
||
}
|
||
return html + renderTextHtml(value.slice(cursor));
|
||
}
|
||
|
||
function serializeEditor(root: HTMLElement) {
|
||
function walk(node: Node): string {
|
||
if (node.nodeType === Node.TEXT_NODE) return node.textContent ?? "";
|
||
if (node.nodeType !== Node.ELEMENT_NODE) return "";
|
||
const element = node as HTMLElement;
|
||
if (element.dataset.pauseMs) return `<pause:${element.dataset.pauseMs}>`;
|
||
if (element.tagName === "BR") return "\n";
|
||
return Array.from(element.childNodes).map(walk).join("");
|
||
}
|
||
return walk(root);
|
||
}
|
||
|
||
function visibleText(value: string) {
|
||
return value.replace(new RegExp(pausePattern.source, "g"), "");
|
||
}
|
||
|
||
function countTypos(value: string) {
|
||
const plain = visibleText(value);
|
||
return typoRules.reduce((count, rule) => count + plain.split(rule.wrong).length - 1, 0);
|
||
}
|
||
|
||
export default function TtsPage() {
|
||
const [config, setConfig] = useState<TtsConfig | null>(null);
|
||
const [voices, setVoices] = useState<Voice[]>([]);
|
||
const [categories, setCategories] = useState<string[]>([]);
|
||
const [facets, setFacets] = useState<Record<string, string[]>>({});
|
||
const [text, setText] = useState("");
|
||
const [voiceId, setVoiceId] = useState("");
|
||
const [category, setCategory] = useState("");
|
||
const [filters, setFilters] = useState<Record<string, string>>({});
|
||
const [search, setSearch] = useState("");
|
||
const [speed, setSpeed] = useState(1);
|
||
const [erhuaStrength, setErhuaStrength] = useState(2);
|
||
const [format, setFormat] = useState("wav");
|
||
const [result, setResult] = useState<Result | null>(null);
|
||
const [loading, setLoading] = useState(true);
|
||
const [busy, setBusy] = useState(false);
|
||
const [helperBusy, setHelperBusy] = useState(false);
|
||
const [error, setError] = useState("");
|
||
const [loadedAt, setLoadedAt] = useState<Date | null>(null);
|
||
const [pauseOpen, setPauseOpen] = useState(false);
|
||
const [customPause, setCustomPause] = useState("3500");
|
||
const [selectedPause, setSelectedPause] = useState<SelectedPause | null>(null);
|
||
const [pauseError, setPauseError] = useState("");
|
||
const [correctionMessage, setCorrectionMessage] = useState("");
|
||
const [helperMessage, setHelperMessage] = useState("");
|
||
const [myVoice, setMyVoice] = useState<{ name: string; dataUrl: string } | null>(null);
|
||
const [recording, setRecording] = useState(false);
|
||
const textareaRef = useRef<HTMLDivElement>(null);
|
||
const fileInputRef = useRef<HTMLInputElement>(null);
|
||
const recorderRef = useRef<MediaRecorder | null>(null);
|
||
const voiceFileRef = useRef<HTMLInputElement>(null);
|
||
|
||
useEffect(() => {
|
||
let cancelled = false;
|
||
async function load() {
|
||
setLoading(true);
|
||
setError("");
|
||
try {
|
||
const [configResponse, voicesResponse] = await Promise.all([
|
||
fetch(apiUrl("/v1/audio/config")),
|
||
fetch(apiUrl("/v1/audio/voices")),
|
||
]);
|
||
if (!configResponse.ok || !voicesResponse.ok) throw new Error("TTS 服务暂时不可用，请稍后重试");
|
||
const nextConfig: TtsConfig = await configResponse.json();
|
||
const nextVoices: VoiceResponse = await voicesResponse.json();
|
||
if (cancelled) return;
|
||
const nextList = nextVoices.voices ?? [];
|
||
setConfig(nextConfig);
|
||
setVoices(nextList);
|
||
setCategories(nextVoices.categories ?? Array.from(new Set(nextList.map((item) => item.category).filter(Boolean) as string[])));
|
||
setFacets(nextVoices.facets ?? {});
|
||
setVoiceId((current) => current || nextList[0]?.id || "");
|
||
setFormat(nextConfig.default_format ?? nextConfig.formats?.[0] ?? "wav");
|
||
setSpeed(nextConfig.speed?.default ?? 1);
|
||
setLoadedAt(new Date());
|
||
} catch (loadError) {
|
||
if (!cancelled) setError(loadError instanceof Error ? loadError.message : "无法连接 TTS 服务");
|
||
} finally {
|
||
if (!cancelled) setLoading(false);
|
||
}
|
||
}
|
||
void load();
|
||
return () => { cancelled = true; };
|
||
}, []);
|
||
|
||
useEffect(() => () => {
|
||
if (result?.url) URL.revokeObjectURL(result.url);
|
||
}, [result?.url]);
|
||
|
||
useEffect(() => {
|
||
if (textareaRef.current && !textareaRef.current.innerHTML) textareaRef.current.innerHTML = renderEditorHtml(text);
|
||
}, [text]);
|
||
|
||
const selectedVoice = voices.find((item) => item.id === voiceId) ?? null;
|
||
const filteredVoices = useMemo(() => {
|
||
const query = search.trim().toLocaleLowerCase();
|
||
return voices.filter((voice) => {
|
||
if (category && voice.category !== category) return false;
|
||
if (query && ![voice.name, voice.id, voice.desc, voice.gender, voice.genre, voice.accent, voice.style].some((value) => value?.toLocaleLowerCase().includes(query))) return false;
|
||
return Object.entries(filters).every(([key, value]) => !value || voice[key as keyof Voice] === value || voice.styles?.includes(value));
|
||
});
|
||
}, [category, filters, search, voices]);
|
||
|
||
useEffect(() => {
|
||
if (filteredVoices.length && !filteredVoices.some((item) => item.id === voiceId)) setVoiceId(filteredVoices[0].id);
|
||
}, [filteredVoices, voiceId]);
|
||
|
||
function setEditorText(nextText: string, caretOffset?: number) {
|
||
const next = normalizeText(nextText);
|
||
setText(next);
|
||
if (textareaRef.current) textareaRef.current.innerHTML = renderEditorHtml(next);
|
||
if (caretOffset !== undefined) {
|
||
window.requestAnimationFrame(() => setCaretAtRawOffset(caretOffset));
|
||
}
|
||
}
|
||
|
||
function setCaretAtRawOffset(offset: number) {
|
||
const root = textareaRef.current;
|
||
if (!root) return;
|
||
const selection = window.getSelection();
|
||
if (!selection) return;
|
||
const range = document.createRange();
|
||
let remaining = Math.max(0, offset);
|
||
let placed = false;
|
||
function visit(node: Node): void {
|
||
if (placed) return;
|
||
if (node.nodeType === Node.TEXT_NODE) {
|
||
const length = node.textContent?.length ?? 0;
|
||
if (remaining <= length) {
|
||
range.setStart(node, remaining);
|
||
range.collapse(true);
|
||
placed = true;
|
||
} else remaining -= length;
|
||
return;
|
||
}
|
||
if (node.nodeType === Node.ELEMENT_NODE && (node as HTMLElement).dataset.pauseMs) {
|
||
const markerLength = `<pause:${(node as HTMLElement).dataset.pauseMs}>`.length;
|
||
if (remaining <= markerLength) {
|
||
range.setStartAfter(node);
|
||
range.collapse(true);
|
||
placed = true;
|
||
} else remaining -= markerLength;
|
||
return;
|
||
}
|
||
Array.from(node.childNodes).forEach(visit);
|
||
}
|
||
visit(root);
|
||
if (!placed) {
|
||
range.selectNodeContents(root);
|
||
range.collapse(false);
|
||
}
|
||
selection.removeAllRanges();
|
||
selection.addRange(range);
|
||
root.focus();
|
||
}
|
||
|
||
function getSelectionOffset() {
|
||
const root = textareaRef.current;
|
||
const currentSelection = window.getSelection();
|
||
if (!root || !currentSelection?.anchorNode || !root.contains(currentSelection.anchorNode)) return text.length;
|
||
const selection = currentSelection;
|
||
let result = 0;
|
||
let reached = false;
|
||
function walk(node: Node): void {
|
||
if (reached) return;
|
||
if (node === selection.anchorNode) {
|
||
result += selection.anchorOffset;
|
||
reached = true;
|
||
return;
|
||
}
|
||
if (node.nodeType === Node.ELEMENT_NODE && (node as HTMLElement).dataset.pauseMs) {
|
||
result += `<pause:${(node as HTMLElement).dataset.pauseMs}>`.length;
|
||
return;
|
||
}
|
||
Array.from(node.childNodes).forEach((child) => {
|
||
if (!reached) walk(child);
|
||
});
|
||
}
|
||
walk(root);
|
||
return reached ? result : text.length;
|
||
}
|
||
|
||
function insertPause(ms: number) {
|
||
const marker = `<pause:${ms}>`;
|
||
const start = getSelectionOffset();
|
||
const next = `${text.slice(0, start)}${marker}${text.slice(start)}`;
|
||
setEditorText(next, start + marker.length);
|
||
setSelectedPause({ start, ms });
|
||
}
|
||
|
||
function choosePause(ms: number) {
|
||
if (selectedPause) {
|
||
const oldMarker = `<pause:${selectedPause.ms}>`;
|
||
if (text.slice(selectedPause.start, selectedPause.start + oldMarker.length) === oldMarker) {
|
||
const marker = `<pause:${ms}>`;
|
||
setEditorText(`${text.slice(0, selectedPause.start)}${marker}${text.slice(selectedPause.start + oldMarker.length)}`, selectedPause.start + marker.length);
|
||
setSelectedPause({ start: selectedPause.start, ms });
|
||
return;
|
||
}
|
||
}
|
||
insertPause(ms);
|
||
}
|
||
|
||
function removeSelectedPause() {
|
||
if (!selectedPause) return;
|
||
const marker = `<pause:${selectedPause.ms}>`;
|
||
if (text.slice(selectedPause.start, selectedPause.start + marker.length) === marker) {
|
||
setEditorText(`${text.slice(0, selectedPause.start)}${text.slice(selectedPause.start + marker.length)}`, selectedPause.start);
|
||
}
|
||
setSelectedPause(null);
|
||
setPauseOpen(false);
|
||
}
|
||
|
||
function applyCustomPause() {
|
||
const ms = Number(customPause);
|
||
const min = config?.pause?.min_ms ?? 50;
|
||
const max = config?.pause?.max_ms ?? 10000;
|
||
if (!Number.isInteger(ms) || ms < min || ms > max) {
|
||
setPauseError(`请输入 ${min}–${max} ms 之间的整数`);
|
||
return;
|
||
}
|
||
setPauseError("");
|
||
choosePause(ms);
|
||
}
|
||
|
||
function applySymbolPauses() {
|
||
const next = text.replace(/([，。！？])(?!<pause:\d+>)/g, (match) => `${match}<pause:${symbolPauses[match]}>`);
|
||
if (next === text) setCorrectionMessage("当前标点已经完成停顿处理");
|
||
else setEditorText(next, next.length);
|
||
}
|
||
|
||
function applyParagraphPauses() {
|
||
const next = text.replace(/\n+(?!<pause:\d+>)/g, (match) => `${match}<pause:2000>`);
|
||
if (next === text) setCorrectionMessage("请先在文本中加入换行，再使用段落停顿");
|
||
else setEditorText(next, next.length);
|
||
}
|
||
|
||
function autoSegment() {
|
||
const next = text.replace(/([。！？])(?!<pause:\d+>|\n)/g, "$1\n");
|
||
if (next === text) setHelperMessage("当前文本已经完成基础分段");
|
||
else {
|
||
setEditorText(next, next.length);
|
||
setHelperMessage("已按句号、问号和感叹号完成自动分段");
|
||
}
|
||
}
|
||
|
||
function handleEditorInput() {
|
||
if (!textareaRef.current) return;
|
||
setText(normalizeText(serializeEditor(textareaRef.current)));
|
||
setSelectedPause(null);
|
||
setCorrectionMessage("");
|
||
}
|
||
|
||
function handleEditorClick(event: { target: EventTarget | null }) {
|
||
const target = event.target instanceof HTMLElement ? event.target : null;
|
||
const pause = target?.closest<HTMLElement>("[data-pause-ms]");
|
||
if (pause) {
|
||
const start = getElementRawOffset(pause);
|
||
setSelectedPause({ start, ms: Number(pause.dataset.pauseMs) });
|
||
setPauseOpen(true);
|
||
setCustomPause(pause.dataset.pauseMs ?? "1000");
|
||
return;
|
||
}
|
||
const typo = target?.closest<HTMLElement>("[data-typo-right]");
|
||
if (typo) {
|
||
const start = getElementRawOffset(typo);
|
||
const wrong = typo.textContent ?? "";
|
||
const right = typo.dataset.typoRight ?? wrong;
|
||
setEditorText(`${text.slice(0, start)}${right}${text.slice(start + wrong.length)}`, start + right.length);
|
||
setCorrectionMessage(`已将「${wrong}」替换为「${right}」`);
|
||
}
|
||
}
|
||
|
||
function getElementRawOffset(element: HTMLElement) {
|
||
const root = textareaRef.current;
|
||
if (!root) return 0;
|
||
let offset = 0;
|
||
let found = false;
|
||
function walk(node: Node): void {
|
||
if (found) return;
|
||
if (node === element) {
|
||
found = true;
|
||
return;
|
||
}
|
||
if (node.nodeType === Node.TEXT_NODE) offset += node.textContent?.length ?? 0;
|
||
else if (node.nodeType === Node.ELEMENT_NODE && (node as HTMLElement).dataset.pauseMs) offset += `<pause:${(node as HTMLElement).dataset.pauseMs}>`.length;
|
||
else Array.from(node.childNodes).forEach((child) => { if (!found) walk(child); });
|
||
}
|
||
walk(root);
|
||
return offset;
|
||
}
|
||
|
||
async function createSpeech() {
|
||
setError("");
|
||
if (!text.trim()) return setError("请输入需要合成的文本");
|
||
if (!myVoice && !voiceId) return setError("请选择一个音色");
|
||
setBusy(true);
|
||
try {
|
||
let blob: Blob;
|
||
if (myVoice) {
|
||
const response = await fetch("/tts-clone", {
|
||
method: "POST",
|
||
headers: { "Content-Type": "application/json" },
|
||
body: JSON.stringify({ text: normalizeText(text), reference_audio: myVoice.dataUrl, response_format: format, speed }),
|
||
});
|
||
if (!response.ok) {
|
||
const body = await response.json().catch(() => ({}));
|
||
throw new Error(body.error?.message ?? `真人语音合成失败（${response.status}）`);
|
||
}
|
||
blob = await response.blob();
|
||
} else {
|
||
const response = await fetch(apiUrl("/v1/audio/speech"), {
|
||
method: "POST",
|
||
headers: { "Content-Type": "application/json" },
|
||
body: JSON.stringify({ input: normalizeText(text), voice: voiceId, response_format: format, speed }),
|
||
});
|
||
if (!response.ok) {
|
||
const body = await response.json().catch(() => ({}));
|
||
throw new Error(body.error?.message ?? `合成失败（${response.status}）`);
|
||
}
|
||
blob = await response.blob();
|
||
}
|
||
const url = URL.createObjectURL(blob);
|
||
setResult({ url, format, fileName: `kaotings-tts-${new Date().toISOString().replace(/[:.]/g, "-")}.${format}` });
|
||
} catch (speechError) {
|
||
setError(speechError instanceof Error ? speechError.message : "语音合成失败");
|
||
} finally {
|
||
setBusy(false);
|
||
}
|
||
}
|
||
|
||
async function startRecording() {
|
||
if (recording) return;
|
||
try {
|
||
const stream = await navigator.mediaDevices.getUserMedia({ audio: true });
|
||
const preferred = ["audio/webm", "audio/mp4", "audio/ogg"].find((mime) => MediaRecorder.isTypeSupported(mime));
|
||
const recorder = new MediaRecorder(stream, preferred ? { mimeType: preferred } : undefined);
|
||
const chunks: Blob[] = [];
|
||
recorder.ondataavailable = (event) => { if (event.data.size > 0) chunks.push(event.data); };
|
||
recorder.onstop = () => {
|
||
stream.getTracks().forEach((track) => track.stop());
|
||
const blob = new Blob(chunks, { type: recorder.mimeType || "audio/webm" });
|
||
const reader = new FileReader();
|
||
reader.onload = () => {
|
||
setMyVoice({ dataUrl: String(reader.result), name: `我的声音 ${new Date().toLocaleTimeString("zh-CN", { hour12: false })}` });
|
||
setRecording(false);
|
||
};
|
||
reader.onerror = () => setRecording(false);
|
||
reader.readAsDataURL(blob);
|
||
};
|
||
recorderRef.current = recorder;
|
||
recorder.start();
|
||
setRecording(true);
|
||
} catch {
|
||
setHelperMessage("无法访问麦克风，请允许浏览器使用麦克风，或直接上传录好的音频文件");
|
||
}
|
||
}
|
||
|
||
function stopRecording() {
|
||
if (recorderRef.current && recorderRef.current.state !== "inactive") recorderRef.current.stop();
|
||
}
|
||
|
||
function handleVoiceFile(file: File | undefined) {
|
||
if (!file) return;
|
||
const isAudio = file.type.startsWith("audio/") || /\.(wav|mp3|m4a|ogg|webm|aac)$/i.test(file.name);
|
||
if (!isAudio) {
|
||
setHelperMessage("请上传音频文件（wav / mp3 / m4a 等）");
|
||
return;
|
||
}
|
||
if (file.size > 8 * 1024 * 1024) {
|
||
setHelperMessage("声音样本不能超过 8 MB，建议 10–60 秒清晰人声");
|
||
return;
|
||
}
|
||
const reader = new FileReader();
|
||
reader.onload = () => setMyVoice({ dataUrl: String(reader.result), name: file.name });
|
||
reader.readAsDataURL(file);
|
||
}
|
||
|
||
function clearMyVoice() {
|
||
setMyVoice(null);
|
||
}
|
||
|
||
async function handleOcr(file: File | undefined) {
|
||
if (!file) return;
|
||
if (!file.type.startsWith("image/")) {
|
||
setHelperMessage("请选择图片文件");
|
||
return;
|
||
}
|
||
if (file.size > 12 * 1024 * 1024) {
|
||
setHelperMessage("图片不能超过 12 MB");
|
||
return;
|
||
}
|
||
setHelperBusy(true);
|
||
setHelperMessage(`正在识别 ${file.name}…`);
|
||
try {
|
||
const imageDataUrl = await new Promise<string>((resolve, reject) => {
|
||
const reader = new FileReader();
|
||
reader.addEventListener("load", () => resolve(String(reader.result)));
|
||
reader.addEventListener("error", () => reject(new Error("图片读取失败")));
|
||
reader.readAsDataURL(file);
|
||
});
|
||
const response = await fetch("/api/v1/ocr", {
|
||
method: "POST",
|
||
credentials: "include",
|
||
headers: { "Content-Type": "application/json" },
|
||
body: JSON.stringify({ image_data_url: imageDataUrl, filename: file.name }),
|
||
});
|
||
if (response.status === 401) throw new Error("请先登录后使用图片文字识别");
|
||
if (!response.ok) {
|
||
const body = await response.json().catch(() => ({}));
|
||
throw new Error(body.error?.message ?? "OCR 服务暂未连接");
|
||
}
|
||
const data = await response.json();
|
||
if (typeof data.text !== "string" || !data.text.trim()) throw new Error("没有识别到可用文字");
|
||
setEditorText(`${text ? `${text}\n` : ""}${data.text}`, text.length + data.text.length + (text ? 1 : 0));
|
||
setHelperMessage("图片文字已填入上方编辑区，可继续校对");
|
||
} catch (ocrError) {
|
||
setHelperMessage(ocrError instanceof Error ? ocrError.message : "图片识别失败，请稍后重试");
|
||
} finally {
|
||
setHelperBusy(false);
|
||
if (fileInputRef.current) fileInputRef.current.value = "";
|
||
}
|
||
}
|
||
|
||
function checkTypos() {
|
||
const count = countTypos(text);
|
||
setCorrectionMessage(count ? `发现 ${count} 处疑似错别字，点击红色波浪线即可替换` : "暂未发现内置词库中的疑似错别字");
|
||
}
|
||
|
||
const formatOptions = config?.formats?.length ? config.formats : ["wav", "mp3", "flac", "opus", "aac", "pcm"];
|
||
const minSpeed = config?.speed?.min ?? 0.25;
|
||
const maxSpeed = config?.speed?.max ?? 4;
|
||
const minPause = config?.pause?.min_ms ?? 50;
|
||
const maxPause = config?.pause?.max_ms ?? 10000;
|
||
|
||
return <div className="container-shell py-16 sm:py-24">
|
||
<div className="flex flex-col justify-between gap-8 lg:flex-row lg:items-end">
|
||
<SectionHeading eyebrow="AI Lab / Text to Speech" title="把文字变成声音" description="输入文案、选择音色，剩下的交给声音工作台。停顿会以直观胶囊显示，技术标记只在提交时保留。" tone="cyan" />
|
||
<span className="rounded-full border border-cyan/30 bg-cyan/8 px-3 py-1 text-xs font-semibold text-cyan">{config?.model ?? "Qwen3-TTS"}</span>
|
||
</div>
|
||
|
||
<div className="mt-10 grid gap-6 lg:grid-cols-[1.2fr_0.8fr]">
|
||
<section className="rounded-2xl border border-line bg-panel/60 p-5 sm:p-7" aria-labelledby="tts-input-title">
|
||
<div className="flex items-center justify-between gap-4">
|
||
<div><p className="text-xs font-semibold uppercase tracking-[0.18em] text-subtle">01 / Input</p><h2 className="mt-2 text-xl font-semibold" id="tts-input-title">输入文本</h2></div>
|
||
<span className="text-xs text-subtle">{visibleText(text).length} 字</span>
|
||
</div>
|
||
<div
|
||
aria-label="配音文本编辑区"
|
||
className={`tts-editor mt-6 min-h-64 leading-7 ${!text ? "is-empty" : ""}`}
|
||
contentEditable
|
||
onClick={handleEditorClick}
|
||
onInput={handleEditorInput}
|
||
onKeyDown={(event) => {
|
||
if (event.key === "Enter") {
|
||
event.preventDefault();
|
||
document.execCommand("insertLineBreak");
|
||
}
|
||
}}
|
||
onPaste={(event) => {
|
||
event.preventDefault();
|
||
document.execCommand("insertText", false, event.clipboardData.getData("text/plain"));
|
||
}}
|
||
ref={textareaRef}
|
||
aria-multiline="true"
|
||
role="textbox"
|
||
spellCheck
|
||
suppressContentEditableWarning
|
||
/>
|
||
<div className="mt-5 border-t border-line pt-5">
|
||
<div className="flex flex-wrap items-center gap-2">
|
||
<span className="mr-1 text-sm text-muted">输入停顿</span>
|
||
<button aria-expanded={pauseOpen} className="button-ghost min-h-9 px-3 py-1.5 text-xs" onClick={() => setPauseOpen((open) => !open)} type="button">⏸ 停顿时长 <span aria-hidden="true">⌄</span></button>
|
||
<button className="tool-button min-h-9" onClick={applySymbolPauses} type="button">， 符号停顿</button>
|
||
<button className="tool-button min-h-9" onClick={applyParagraphPauses} type="button">¶ 段落停顿</button>
|
||
<button className="tool-button min-h-9" onClick={checkTypos} type="button">✓ 错别字</button>
|
||
<button className="button-primary" disabled={busy || loading} onClick={() => void createSpeech()} type="button"><BrandIcon />{busy ? "合成中…" : "生成语音"}<ArrowRight /></button>
|
||
</div>
|
||
{pauseOpen ? <div className="pause-panel mt-4">
|
||
<div className="flex flex-wrap items-center gap-2">
|
||
{pauseOptions.map((ms) => <button className={selectedPause?.ms === ms ? "pause-option is-selected" : "pause-option"} key={ms} onClick={() => choosePause(ms)} type="button">{ms / 1000}s</button>)}
|
||
<span className="text-xs text-subtle">{selectedPause ? "正在编辑已选胶囊" : "点击选项插入到光标处"}</span>
|
||
</div>
|
||
<div className="mt-4 flex flex-wrap items-center gap-3 text-sm text-muted">
|
||
<label htmlFor="custom-pause">自定义</label>
|
||
<input className="pause-input" id="custom-pause" max={maxPause} min={minPause} onChange={(event) => { setCustomPause(event.target.value); setPauseError(""); }} type="number" value={customPause} />
|
||
<span>ms</span>
|
||
<input aria-label="自定义停顿时长" className="pause-range" max={maxPause} min={minPause} onChange={(event) => { setCustomPause(event.target.value); setPauseError(""); }} step="50" type="range" value={Math.min(maxPause, Math.max(minPause, Number(customPause) || minPause))} />
|
||
<button className="button-ghost min-h-9 px-3 py-1.5 text-xs" onClick={applyCustomPause} type="button">应用</button>
|
||
{selectedPause ? <button className="text-xs text-danger hover:underline" onClick={removeSelectedPause} type="button">删除已选胶囊</button> : null}
|
||
</div>
|
||
{pauseError ? <p className="mt-2 text-xs text-danger">{pauseError}</p> : null}
|
||
<p className="mt-3 text-xs leading-5 text-subtle">范围 {minPause}–{maxPause} ms。技术写法已隐藏，点击编辑区里的 ⏸ 胶囊可修改或删除。</p>
|
||
</div> : null}
|
||
<p className="mt-3 text-xs leading-5 text-subtle">点按钮或选时长即可插入停顿，无需手写代码。标点停顿默认：逗号 0.3s，句号 / 问号 / 感叹号 0.8s。</p>
|
||
{correctionMessage ? <p className="mt-3 rounded-lg border border-cyan/20 bg-cyan/5 px-3 py-2 text-xs text-cyan">{correctionMessage}</p> : null}
|
||
</div>
|
||
<p className="mt-5 border-t border-line pt-5 text-sm text-muted">{loadedAt ? `音色清单已更新 · ${voices.length} 个音色` : "正在读取音色清单…"}</p>
|
||
<div className="mt-6 grid gap-4 md:grid-cols-3">
|
||
<button className="assist-card text-left" disabled={helperBusy} onClick={() => fileInputRef.current?.click()} onDragOver={(event) => event.preventDefault()} onDrop={(event) => { event.preventDefault(); void handleOcr(event.dataTransfer.files[0]); }} type="button"><span className="assist-icon">▧</span><span><strong>图片转文字</strong><small>上传或拖入截图、纸稿、PPT，识别结果填入编辑区</small></span><span className="assist-arrow">→</span></button>
|
||
<button className="assist-card text-left" onClick={checkTypos} type="button"><span className="assist-icon">✓</span><span><strong>生成前查错字</strong><small>扫描疑似错字，红色波浪线标出并支持一键替换</small></span><span className="assist-arrow">→</span></button>
|
||
<button className="assist-card text-left" onClick={autoSegment} type="button"><span className="assist-icon">↵</span><span><strong>自动分段</strong><small>按句号、问号和感叹号切分长文案，方便控制节奏</small></span><span className="assist-arrow">→</span></button>
|
||
</div>
|
||
<input accept="image/*" className="hidden" onChange={(event) => void handleOcr(event.target.files?.[0])} ref={fileInputRef} type="file" />
|
||
{helperMessage ? <p className="mt-4 rounded-lg border border-cyan/20 bg-cyan/5 px-3 py-2 text-xs text-cyan">{helperMessage}</p> : null}
|
||
<section aria-label="真人声音配音模式" className={`voice-clone-panel mt-6 ${myVoice ? "is-ready" : ""}`}>
|
||
<div className="flex flex-wrap items-center gap-x-4 gap-y-3">
|
||
<div className="min-w-0 flex-1">
|
||
<p className="font-semibold"><span aria-hidden="true">🎙</span> 真人声音配音模式</p>
|
||
<p className="mt-1 text-xs leading-5 text-muted">录制自己的声音或上传声音样本；就绪后在上方编辑区输入文案，点击「生成语音」即可用自己的声音朗读。</p>
|
||
</div>
|
||
<div className="flex flex-wrap items-center gap-2">
|
||
{recording ? <button className="button-secondary" onClick={stopRecording} type="button">■ 停止录音</button> : <button className="button-secondary" onClick={() => { void startRecording(); }} type="button">● 录制我的声音</button>}
|
||
<button className="button-ghost min-h-9 px-3 py-1.5 text-xs" onClick={() => voiceFileRef.current?.click()} type="button">上传声音</button>
|
||
{myVoice ? <button className="text-xs text-danger hover:underline" onClick={clearMyVoice} type="button">清除样本</button> : null}
|
||
</div>
|
||
</div>
|
||
{myVoice ? <div className="mt-4 flex flex-wrap items-center gap-3">
|
||
<span className="rounded-full border border-cyan/30 bg-cyan/10 px-3 py-1 text-xs font-semibold text-cyan">✓ {myVoice.name} 已就绪，生成时将使用此声音</span>
|
||
<audio className="h-9 w-full max-w-96" controls preload="metadata" src={myVoice.dataUrl} />
|
||
</div> : null}
|
||
{recording ? <p className="mt-3 text-xs text-cyan">● 正在录音…请朗读 10–60 秒清晰人声，完成后点击「停止录音」</p> : null}
|
||
<input accept="audio/*,.wav,.mp3,.m4a" className="hidden" onChange={(event) => { handleVoiceFile(event.target.files?.[0]); event.target.value = ""; }} ref={voiceFileRef} type="file" />
|
||
</section>
|
||
</section>
|
||
|
||
<aside className="space-y-6">
|
||
<section className="rounded-2xl border border-line bg-panel/60 p-5 sm:p-7" aria-labelledby="tts-settings-title">
|
||
<p className="text-xs font-semibold uppercase tracking-[0.18em] text-subtle">02 / Voice & Parameters</p><h2 className="mt-2 text-xl font-semibold" id="tts-settings-title">声音参数</h2>
|
||
<div className="mt-6 space-y-5">
|
||
<div><label className="field-label" htmlFor="voice-search">搜索音色</label><input className="field-input" id="voice-search" onChange={(event) => setSearch(event.target.value)} placeholder="按名称、描述、方言或风格搜索" value={search} /></div>
|
||
<div className="grid gap-4 sm:grid-cols-2"><div><label className="field-label" htmlFor="voice-category">分类</label><select className="field-input" id="voice-category" onChange={(event) => setCategory(event.target.value)} value={category}><option value="">全部分类</option>{categories.map((item) => <option key={item} value={item}>{item}</option>)}</select></div>{Object.entries(facets).map(([key, values]) => <div key={key}><label className="field-label" htmlFor={`facet-${key}`}>{({ gender: "性别", genre: "类型", accent: "方言", style: "风格" } as Record<string, string>)[key] ?? key}</label><select className="field-input" id={`facet-${key}`} onChange={(event) => setFilters((current) => ({ ...current, [key]: event.target.value }))} value={filters[key] ?? ""}><option value="">全部</option>{values.map((item) => <option key={item} value={item}>{item}</option>)}</select></div>)}</div>
|
||
<div><label className="field-label" htmlFor="voice">音色 <span className="font-normal text-subtle">({filteredVoices.length} 个匹配)</span></label><select className="field-input" id="voice" onChange={(event) => setVoiceId(event.target.value)} value={voiceId}><option value="">选择音色</option>{categories.map((group) => { const groupVoices = filteredVoices.filter((item) => item.category === group); return groupVoices.length ? <optgroup key={group} label={group}>{groupVoices.map((item) => <option key={item.id} value={item.id}>{item.name}{item.accent ? ` · ${item.accent}` : ""}</option>)}</optgroup> : null; })}{filteredVoices.filter((item) => !item.category || !categories.includes(item.category)).map((item) => <option key={item.id} value={item.id}>{item.name}</option>)}</select></div>
|
||
{selectedVoice ? <div className="rounded-xl border border-cyan/20 bg-cyan/5 p-4"><p className="font-semibold text-copy">{selectedVoice.name}</p><p className="mt-1 text-sm leading-6 text-muted">{selectedVoice.desc || "暂无音色描述"}</p><div className="mt-3 flex flex-wrap gap-2 text-xs text-cyan">{[selectedVoice.gender, selectedVoice.genre, selectedVoice.accent, selectedVoice.style].filter(Boolean).map((tag) => <span className="rounded-full border border-cyan/20 px-2 py-1" key={tag}>{tag}</span>)}</div>{selectedVoice.preview ? <audio className="mt-4 w-full" controls preload="none" src={apiUrl(selectedVoice.preview)} /> : null}</div> : null}
|
||
<div><label className="field-label" htmlFor="speed">语速 <span className="font-normal text-subtle">{speed.toFixed(2)}x</span></label><input className="w-full accent-cyan" id="speed" max={maxSpeed} min={minSpeed} onChange={(event) => setSpeed(Number(event.target.value))} step="0.05" type="range" value={speed} /><div className="mt-1 flex justify-between text-xs text-subtle"><span>{minSpeed}x</span><span>{maxSpeed}x</span></div></div>
|
||
<div><label className="field-label" htmlFor="erhua">儿化音强度 <span className="font-normal text-cyan">{erhuaOptions[erhuaStrength]}</span></label><input className="w-full accent-cyan" id="erhua" max="3" min="0" onChange={(event) => setErhuaStrength(Number(event.target.value))} step="1" type="range" value={erhuaStrength} /><div className="mt-1 flex justify-between text-xs text-subtle"><span>关闭</span><span>弱</span><span>标准</span><span>强</span></div><p className="mt-2 text-xs leading-5 text-subtle">真人音色儿化偏重时，可调到「弱」或「关闭」。</p></div>
|
||
<div><label className="field-label" htmlFor="format">输出格式</label><select className="field-input" id="format" onChange={(event) => setFormat(event.target.value)} value={format}>{formatOptions.map((item) => <option key={item} value={item}>{item.toUpperCase()}</option>)}</select></div>
|
||
</div>
|
||
</section>
|
||
<StatusCard title="服务状态" description={error || (loading ? "正在连接 TTS 服务…" : `已连接 · ${voices.length} 个音色可用`)} tone={error ? "warning" : "info"} />
|
||
</aside>
|
||
</div>
|
||
|
||
{error ? <p className="mt-6 rounded-xl border border-danger/30 bg-danger/10 px-4 py-3 text-sm leading-6 text-danger" role="alert">{error}</p> : null}
|
||
{result ? <section className="mt-8 rounded-2xl border border-success/30 bg-success/5 p-5 sm:p-7" aria-labelledby="tts-result-title"><div className="flex items-start justify-between gap-4"><div><p className="text-xs font-semibold uppercase tracking-[0.18em] text-success">04 / Result</p><h2 className="mt-2 text-xl font-semibold" id="tts-result-title">生成完成</h2><p className="mt-2 break-all text-sm text-muted">{result.fileName}</p></div><WaveIcon className="h-8 w-8 shrink-0 text-success" /></div><audio autoPlay className="mt-5 w-full" controls src={result.url} /><div className="mt-4 flex justify-end"><a className="button-secondary" download={result.fileName} href={result.url}><BrandIcon />下载音频</a></div></section> : <section className="mt-8 rounded-2xl border border-line bg-panel/40 p-5 sm:p-7"><div className="flex min-h-28 flex-col items-center justify-center text-center"><span className="text-cyan"><WaveIcon className="h-9 w-9" /></span><p className="mt-3 font-semibold">生成结果会显示在这里</p><p className="mt-1 text-sm text-muted">选择音色并输入文本后，即可试听和下载。</p></div></section>}
|
||
|
||
<details className="mt-6 rounded-xl border border-line bg-panel/30 px-4 py-3 text-xs text-subtle"><summary className="cursor-pointer font-semibold text-muted">高级：停顿技术写法</summary><p className="mt-3 leading-6">服务端仍兼容 <code><pause:毫秒></code>、<code>[pause:毫秒]</code> 和 SSML break 写法；普通使用无需接触这些标记，编辑区里的停顿胶囊会在提交时自动转换。</p></details>
|
||
</div>;
|
||
}
|