This commit is contained in:
熊潇 2025-06-23 10:38:10 +08:00
commit 8bc5ba1892
8 changed files with 5696 additions and 0 deletions

3
.gitignore vendored Normal file
View File

@ -0,0 +1,3 @@
node_modules
videos

3
.gitmodules vendored Normal file
View File

@ -0,0 +1,3 @@
[submodule "packages/video-tools"]
path = packages/video-tools
url = git@git.xiongxiao.me:kevisual/video-tools.git

23
package.json Normal file
View File

@ -0,0 +1,23 @@
{
"name": "@kevisual/batch-asr",
"version": "0.0.1",
"description": "",
"main": "index.js",
"scripts": {
"test": "echo \"Error: no test specified\" && exit 1"
},
"keywords": [],
"author": "abearxiong <xiongxiao@xiongxiao.me> (https://www.xiongxiao.me)",
"license": "MIT",
"packageManager": "pnpm@10.12.1",
"type": "module",
"devDependencies": {
"@kevisual/types": "^0.0.10",
"@kevisual/video-tools": "workspace:*",
"@types/node": "^24.0.3"
},
"dependencies": {
"eventemitter3": "^5.0.1",
"natural": "^8.1.0"
}
}

1
packages/video-tools Submodule

@ -0,0 +1 @@
Subproject commit 767e436eb8ba3d38bd68e90e18203b039662dce1

5512
pnpm-lock.yaml generated Normal file

File diff suppressed because it is too large Load Diff

6
pnpm-workspace.yaml Normal file
View File

@ -0,0 +1,6 @@
packages:
- packages/*
onlyBuiltDependencies:
- esbuild
- msw

130
src/asr.ts Normal file
View File

@ -0,0 +1,130 @@
import { VideoWS, VideoWsResult } from '@kevisual/video-tools/src/asr/provider/funasr/ws.ts';
import net from 'node:net';
import path from 'node:path';
import fs from 'node:fs';
import natural from 'natural';
import { EventEmitter } from 'eventemitter3';
const sleep = (ms: number) => new Promise((resolve) => setTimeout(resolve, ms));
const videoTestPath = path.join(process.cwd(), 'videos/my_speech_text.wav');
const audioTestPath = path.join(process.cwd(), 'videos/test.mp3');
const name = 'output-1746007775571.mp3';
const videoTestPath2 = path.join(process.cwd(), 'build', name);
const emitter = new EventEmitter();
const url = 'https://funasr.xiongxiao.me';
const local = 'wss://192.168.31.220:10095';
const ws = new VideoWS({
url: url,
mode: '2pass',
wsOptions: {
rejectUnauthorized: false,
},
emitter,
wav_format: 'wav',
onConnect: async () => {
console.log('WebSocket connected');
},
});
type BatchSenderOptions = {
vws: VideoWS;
files: string[];
matchText?: string;
emitter?: EventEmitter;
};
class BatchSenderFiles {
files: string[];
vws: VideoWS;
emitter: EventEmitter;
constructor({ vws, files, matchText, emitter }: BatchSenderOptions) {
this.files = files;
this.vws = vws;
this.emitter = emitter || vws.emitter;
}
async init() {
const isConnected = await this.vws.isConnected();
if (!isConnected) {
console.error('链接失败:', isConnected);
}
this.send();
}
waitOne() {
return new Promise((resolve) => {
this.vws.emitter.once('result', (data) => {
resolve(data);
});
});
}
async checkAudioFile(file: string) {
const stats = fs.statSync(file);
if (!stats.isFile()) {
throw new Error(`File not found: ${file}`);
}
const ext = path.extname(file).toLowerCase();
const validExtensions = ['.wav', '.mp3', '.flac', '.ogg', '.aac'];
if (!validExtensions.includes(ext)) {
throw new Error(`Invalid file type: ${ext}. Supported types are: ${validExtensions.join(', ')}`);
}
const fileSize = stats.size;
if (fileSize === 0) {
throw new Error(`File is empty: ${file}`);
}
const maxSize = 100 * 1024 * 1024; // 100 MB
if (fileSize > maxSize) {
throw new Error(`File size exceeds limit: ${fileSize} bytes. Maximum allowed size is ${maxSize} bytes.`);
}
return {
file,
ext,
size: fileSize,
isValid: true,
};
}
async send() {
const textList: { file: string; text: string }[] = [];
for (const file of this.files) {
let wav_format = 'wav';
try {
const ck = await this.checkAudioFile(file);
if (ck.ext !== '.wav') {
wav_format = ck.ext.replace('.', '');
}
} catch (error) {
console.error('Error checking file:', error);
continue;
}
const data = fs.readFileSync(file);
const wait = this.waitOne();
await this.vws.sendBuffer(data, { wav_format });
await sleep(1000);
console.log('File sent:', file);
const result: VideoWsResult = (await wait) as any;
console.log('Result:', result.text);
textList.push({ file, text: result.text });
console.log('----------------------');
}
this.emitter.emit('send-done', textList);
}
}
const batchSender = new BatchSenderFiles({
vws: ws,
// files: [audioTestPath],
files: [videoTestPath, audioTestPath],
});
batchSender.init();
batchSender.emitter.on('send-done', (data) => {
const matchText = '在一无所知中,梦里的一天结束了一个新的轮回,便会开始。';
const textList = data as { file: string; text: string }[];
for (const item of textList) {
const getText = item.text || '';
const distance = natural.JaroWinklerDistance(getText, matchText);
console.log(`File: ${item.file}, \nText: ${item.text}\nDistance: ${distance}`);
}
// console.log('Batch processing done:', data);
});
const server = net.createServer((socket) => {
socket.on('data', (data) => {
console.log('data', data);
});
});
server.listen(10096);

18
tsconfig.json Normal file
View File

@ -0,0 +1,18 @@
{
"extends": "@kevisual/types/json/backend.json",
"compilerOptions": {
"baseUrl": ".",
"typeRoots": [
"./node_modules/@types",
"./node_modules/@kevisual"
],
"paths": {
"@/*": [
"src/*"
]
},
},
"include": [
"src/**/*",
],
}