视频字幕 OCR

复制import subprocess
import json
from pathlib import Path

# 使用 PaddleOCR 提取视频帧中的硬字幕
# 前置条件：pip install paddleocr paddlepaddle opencv-python

def extract_subtitles(video_path: str, output_srt: str):
    """从视频中提取硬字幕并生成 SRT 文件"""
    from paddleocr import PaddleOCR
    import cv2
    
    ocr = PaddleOCR(use_angle_cls=True, lang='ch')
    cap = cv2.VideoCapture(video_path)
    fps = cap.get(cv2.CAP_PROP_FPS)
    
    subtitles = []
    frame_idx = 0
    
    while cap.isOpened():
        ret, frame = cap.read()
        if not ret:
            break
        
        # 每 30 帧处理一次（减少重复检测）
        if frame_idx % 30 == 0:
            result = ocr.ocr(frame, cls=True)
            if result and result[0]:
                text = ' '.join([line[1][0] for line in result[0]])
                if text.strip():
                    time_sec = frame_idx / fps
                    subtitles.append((time_sec, text))
        
        frame_idx += 1
    
    cap.release()
    
    # 生成 SRT 格式
    with open(output_srt, 'w', encoding='utf-8') as f:
        for i, (start_time, text) in enumerate(subtitles, 1):
            start = f"{int(start_time//3600):02d}:{int((start_time%3600)//60):02d}:{start_time%60:06.3f}"
            end_time = start_time + 1.0  # 假设每段字幕持续 1 秒
            end = f"{int(end_time//3600):02d}:{int((end_time%3600)//60):02d}:{end_time%60:06.3f}"
            f.write(f"{i}\n{start} --> {end}\n{text}\n\n")
    
    print(f"已提取 {len(subtitles)} 条字幕到 {output_srt}")

# 使用示例
extract_subtitles("input.mp4", "output.srt")

复制package main

import (
	"fmt"
	"os"
	"os/exec"
	"path/filepath"
	"strings"
)

// 使用 Tesseract OCR 从视频帧提取字幕
// 前置条件：安装 tesseract-ocr + 中文语言包

func extractSubtitles(videoPath, outputSrt string) error {
	// 1. 用 ffmpeg 每隔 1 秒截取一帧
	framesDir := filepath.Dir(outputSrt) + "/frames"
	os.MkdirAll(framesDir, 0755)
	
	cmd := exec.Command("ffmpeg", "-i", videoPath, "-vf", "fps=1", framesDir+"/frame_%04d.png")
	if err := cmd.Run(); err != nil {
		return fmt.Errorf("ffmpeg 截帧失败: %w", err)
	}

	// 2. 对每帧进行 OCR
	subtitles := []string{}
	files, _ := filepath.Glob(framesDir + "/frame_*.png")
	
	for _, frame := range files {
		outFile := frame + ".txt"
		cmd := exec.Command("tesseract", frame, outFile[:len(outFile)-4], "-l", "chi_sim+eng", "--psm", "6")
		if err := cmd.Run(); err != nil {
			continue // 跳过 OCR 失败的帧
		}
		
		data, _ := os.ReadFile(outFile)
		text := strings.TrimSpace(string(data))
		if len(text) > 0 {
			subtitles = append(subtitles, text)
		}
	}

	// 3. 生成 SRT（简化版：每段持续 1 秒）
	var srtBuilder strings.Builder
	for i, text := range subtitles {
		start := fmt.Sprintf("%02d:%02d:%02d,000", i/3600, (i%3600)/60, i%60)
		end := fmt.Sprintf("%02d:%02d:%02d,000", (i+1)/3600, ((i+1)%3600)/60, (i+1)%60)
		srtBuilder.WriteString(fmt.Sprintf("%d\n%s --> %s\n%s\n\n", i+1, start, end, text))
	}
	
	os.WriteFile(outputSrt, []byte(srtBuilder.String()), 0644)
	os.RemoveAll(framesDir)
	
	fmt.Printf("提取完成，共 %d 条字幕\n", len(subtitles))
	return nil
}

func main() {
	if err := extractSubtitles("input.mp4", "output.srt"); err != nil {
		fmt.Fprintf(os.Stderr, "错误: %v\n", err)
		os.Exit(1)
	}
}

复制// 使用 Tesseract.js 在浏览器中提取视频字幕
// 前置条件：npm install tesseract.js

const Tesseract = require('tesseract.js');
const { createCanvas, loadImage } = require('canvas');
const fs = require('fs');

async function extractSubtitles(videoPath, outputSrt) {
  // 使用 ffmpeg 将视频转为帧序列（需在 Node.js 环境运行）
  const { execSync } = require('child_process');
  execSync(`ffmpeg -i ${videoPath} -vf fps=1 frames/frame_%04d.png`);

  const frames = fs.readdirSync('frames').filter(f => f.endsWith('.png'));
  const subtitles = [];

  for (let i = 0; i < frames.length; i++) {
    const imagePath = `frames/${frames[i]}`;
    
    // 使用 Tesseract.js 进行 OCR
    const { data } = await Tesseract.recognize(imagePath, 'chi_sim+eng', {
      logger: m => {} // 静默模式
    });

    const text = data.text.trim();
    if (text) {
      const timeSec = i; // 每秒一帧
      subtitles.push({ start: timeSec, text });
    }
  }

  // 生成 SRT 格式
  let srtContent = '';
  subtitles.forEach((sub, idx) => {
    const start = new Date(sub.start * 1000).toISOString().substr(11, 12).replace('.', ',');
    const end = new Date((sub.start + 1) * 1000).toISOString().substr(11, 12).replace('.', ',');
    srtContent += `${idx + 1}\n${start} --> ${end}\n${sub.text}\n\n`;
  });

  fs.writeFileSync(outputSrt, srtContent, 'utf-8');
  console.log(`已提取 ${subtitles.length} 条字幕到 ${outputSrt}`);
  
  // 清理临时帧
  fs.rmSync('frames', { recursive: true, force: true });
}

// 使用示例
extractSubtitles('input.mp4', 'output.srt').catch(console.error);