Files
CRM_Montzar/apps/web/src/components/VoiceToTextButton.vue
T

335 lines
8.2 KiB
Vue
Raw Normal View History

<script setup lang="ts">
import { computed, onBeforeUnmount, ref } from 'vue';
import axios from 'axios';
import { api } from '@/lib/api';
import { toFaDigits } from '@/lib/jalali';
import UiIcon from '@/components/UiIcon.vue';
const props = withDefaults(
defineProps<{
disabled?: boolean;
variant?: 'default' | 'toolbar';
hideError?: boolean;
}>(),
{ variant: 'default', hideError: false },
);
const emit = defineEmits<{
transcript: [text: string];
recording: [active: boolean];
busy: [active: boolean];
error: [message: string];
}>();
const recording = ref(false);
const transcribing = ref(false);
const error = ref('');
const seconds = ref(0);
let stream: MediaStream | null = null;
let ctx: AudioContext | null = null;
let source: MediaStreamAudioSourceNode | null = null;
let processor: ScriptProcessorNode | null = null;
let mute: GainNode | null = null;
let chunks: Float32Array[] = [];
let sampleRate = 16000;
let timer: number | null = null;
let limitTimer: number | null = null;
const MAX_SECONDS = 90;
const busy = computed(() => recording.value || transcribing.value);
const timerLabel = computed(() => {
const m = String(Math.floor(seconds.value / 60)).padStart(2, '0');
const s = String(seconds.value % 60).padStart(2, '0');
return toFaDigits(`${m}:${s}`);
});
const ariaLabel = computed(() => {
if (transcribing.value) return 'در حال تبدیل صوت به متن';
if (recording.value) return 'توقف ضبط و تبدیل به متن';
return 'تبدیل صوت به متن';
});
function setError(message: string) {
error.value = message;
emit('error', message);
}
function setRecording(active: boolean) {
recording.value = active;
emit('recording', active);
emit('busy', busy.value);
}
function setTranscribing(active: boolean) {
transcribing.value = active;
emit('busy', recording.value || active);
}
function stopTimer() {
if (timer != null) {
window.clearInterval(timer);
timer = null;
}
if (limitTimer != null) {
window.clearTimeout(limitTimer);
limitTimer = null;
}
}
function teardownMic() {
stopTimer();
try {
processor?.disconnect();
source?.disconnect();
mute?.disconnect();
} catch {
/* ignore */
}
processor = null;
source = null;
mute = null;
stream?.getTracks().forEach((track) => track.stop());
stream = null;
if (ctx) {
void ctx.close().catch(() => undefined);
ctx = null;
}
}
function mergeChunks(parts: Float32Array[]) {
const length = parts.reduce((sum, part) => sum + part.length, 0);
const out = new Float32Array(length);
let offset = 0;
for (const part of parts) {
out.set(part, offset);
offset += part.length;
}
return out;
}
function encodeWav(samples: Float32Array, rate: number) {
const bytes = samples.length * 2;
const buffer = new ArrayBuffer(44 + bytes);
const view = new DataView(buffer);
const write = (offset: number, text: string) => {
for (let i = 0; i < text.length; i += 1) view.setUint8(offset + i, text.charCodeAt(i));
};
write(0, 'RIFF');
view.setUint32(4, 36 + bytes, true);
write(8, 'WAVE');
write(12, 'fmt ');
view.setUint32(16, 16, true);
view.setUint16(20, 1, true);
view.setUint16(22, 1, true);
view.setUint32(24, rate, true);
view.setUint32(28, rate * 2, true);
view.setUint16(32, 2, true);
view.setUint16(34, 16, true);
write(36, 'data');
view.setUint32(40, bytes, true);
let offset = 44;
for (let i = 0; i < samples.length; i += 1) {
const sample = Math.max(-1, Math.min(1, samples[i] ?? 0));
view.setInt16(offset, sample < 0 ? sample * 0x8000 : sample * 0x7fff, true);
offset += 2;
}
return new Blob([buffer], { type: 'audio/wav' });
}
async function upload(blob: Blob) {
const form = new FormData();
form.append('audio', blob, 'dictation.wav');
const { data } = await api.post<{ text: string }>('/ai/transcribe', form);
return (data.text ?? '').trim();
}
async function finish(keep: boolean) {
const parts = chunks;
chunks = [];
const rate = sampleRate;
teardownMic();
setRecording(false);
if (!keep) return;
const samples = mergeChunks(parts);
if (samples.length < rate * 0.35) {
setError('صدای کافی ضبط نشد. کمی بیشتر صحبت کنید.');
return;
}
setTranscribing(true);
try {
const text = await upload(encodeWav(samples, rate));
if (!text) {
setError('متنی از صدا تشخیص داده نشد.');
return;
}
emit('transcript', text);
} catch (err) {
if (axios.isAxiosError(err)) {
const msg = (err.response?.data as { message?: string | string[] })?.message;
setError(Array.isArray(msg) ? msg.join('، ') : (msg ?? 'تبدیل صوت به متن ناموفق بود'));
} else {
setError('تبدیل صوت به متن ناموفق بود');
}
} finally {
setTranscribing(false);
}
}
async function start() {
setError('');
if (!navigator.mediaDevices?.getUserMedia) {
setError('این مرورگر به میکروفون دسترسی نمی‌دهد.');
return;
}
try {
stream = await navigator.mediaDevices.getUserMedia({
audio: { echoCancellation: true, noiseSuppression: true, channelCount: 1 },
});
const Ctor = window.AudioContext || (window as Window & { webkitAudioContext?: typeof AudioContext }).webkitAudioContext;
if (!Ctor) {
setError('ضبط صدا در این مرورگر پشتیبانی نمی‌شود.');
stream.getTracks().forEach((track) => track.stop());
stream = null;
return;
}
ctx = new Ctor();
await ctx.resume();
sampleRate = ctx.sampleRate || 16000;
source = ctx.createMediaStreamSource(stream);
processor = ctx.createScriptProcessor(4096, 1, 1);
mute = ctx.createGain();
mute.gain.value = 0;
chunks = [];
processor.onaudioprocess = (event) => {
if (!recording.value) return;
chunks.push(new Float32Array(event.inputBuffer.getChannelData(0)));
};
source.connect(processor);
processor.connect(mute);
mute.connect(ctx.destination);
seconds.value = 0;
timer = window.setInterval(() => {
seconds.value += 1;
}, 1000);
limitTimer = window.setTimeout(() => {
void finish(true);
}, MAX_SECONDS * 1000);
setRecording(true);
} catch {
teardownMic();
setError('دسترسی به میکروفون داده نشد.');
}
}
function toggle() {
if (props.disabled || transcribing.value) return;
if (recording.value) {
void finish(true);
return;
}
void start();
}
onBeforeUnmount(() => {
void finish(false);
});
</script>
<template>
<div class="vtt-wrap" :class="variant">
<button
class="vtt"
type="button"
:class="{ on: recording, busy: transcribing }"
:disabled="disabled || transcribing"
:aria-label="ariaLabel"
:title="ariaLabel"
@click="toggle"
>
<UiIcon :name="recording ? 'stop' : 'mic'" :size="variant === 'toolbar' ? 18 : 16" />
<span v-if="variant === 'default' && recording" class="timer">{{ timerLabel }}</span>
<span v-else-if="variant === 'default' && transcribing" class="timer">تبدیل…</span>
</button>
<p v-if="error && !hideError" class="err">{{ error }}</p>
</div>
</template>
<style scoped>
.vtt-wrap {
display: grid;
gap: 0.3rem;
justify-items: start;
}
.vtt {
display: inline-flex;
align-items: center;
gap: 0.35rem;
border: 1px solid #d7e3dc;
background: #f4faf7;
color: #0f766e;
border-radius: 999px;
min-height: 36px;
padding: 0 0.75rem;
font: inherit;
font-size: 0.8rem;
font-weight: 700;
cursor: pointer;
}
.vtt:hover:not(:disabled) {
background: #ecfdf5;
}
.vtt.on {
border-color: #f87171;
background: #fef2f2;
color: #b91c1c;
animation: pulse 1.2s ease-in-out infinite;
}
.vtt.busy:not(.on) {
opacity: 0.72;
cursor: wait;
}
.vtt:disabled {
opacity: 0.45;
cursor: not-allowed;
}
.toolbar .vtt {
width: 38px;
height: 38px;
min-height: 38px;
padding: 0;
border-radius: 12px;
border-color: transparent;
background: #f8fafc;
justify-content: center;
}
.toolbar .vtt.on {
background: #fef2f2;
}
.timer {
font-variant-numeric: tabular-nums;
}
.err {
margin: 0;
color: #b91c1c;
font-size: 0.78rem;
}
@keyframes pulse {
50% {
box-shadow: 0 0 0 5px rgba(248, 113, 113, 0.16);
}
}
</style>