分类
ASRT 后端开发 智能语音技术

Java读取并解析wav格式文件

本文将主要介绍如何使用Java语言读取文件并解析wave格式,并以代码形式进行展开。代码主要包含三个功能类,分别为:Wave、DataParseUtils和Common。

本代码已用于GitHub上开源的ASRT语音识别系统的Java语言SDK项目: https://github.com/nl8590687/ASRT_SDK_Java

class Wave:

public class Wave {
    public short[] samples;
    public byte[] sampleBytes;
    public int sampleRate;
    public int channels;
    public int sampleWidth;

    public Wave(){}

    public Wave(short[] samples, int sampleRate, int channels, int sampleWidth) {
        this.samples = samples;
        this.sampleRate = sampleRate;
        this.channels = channels;
        this.sampleWidth = sampleWidth;
        // short[] 转 byte[]
        this.sampleBytes = this.samplesToBytes(samples);
    }

    public Wave(byte[] sampleBytes, int sampleRate, int channels, int sampleWidth) {
        this.sampleBytes = sampleBytes;
        this.sampleRate = sampleRate;
        this.channels = channels;
        this.sampleWidth = sampleWidth;
        // byte[] 转 short[]
        this.samples = this.bytesToSamples(sampleBytes);
    }

    public boolean deserialize(byte[] wavBytes) {
        try
        {
            byte[] riff = new byte[4];
            byte[] riffSize = new byte[4];
            byte[] waveID = new byte[4];
            byte[] junkID = new byte[4];
            boolean hasjunk = false;
            byte[] junklength = new byte[4];

            byte[] fmtID = new byte[4];
            byte[] cksize = new byte[4];
            int waveType = 0; // 无符号int整数,在获取时需要进行字节转码 (Byte.toUnsignedInt(byte x))
            byte[] channel = new byte[2];
            byte[] sample_rate = new byte[4];
            byte[] bytespersec = new byte[4];
            byte[] blocklen_sample = new byte[2];
            byte[] bitNum = new byte[2];
            byte[] unknown = new byte[2];
            byte[] dataID = new byte[4];  //52
            byte[] dataLength = new byte[4];  //56 个字节

            int p = 0; //模拟流的指针位置

            System.arraycopy(wavBytes, p, riff, 0, 4); // RIFF
            p += 4;

            if (DataParseUtils.convertFoutUnsignLong(riff[3], riff[2], riff[1], riff[0]) != 0x52494646) //0x52494646
            {
                Exception e = new Exception("该文件不是WAVE文件");
                throw e;
            }

            /*if (riff[0]!=82 || riff[1]!=73  || riff[2]!=70  || riff[3]!=70) //0x52494646
            {
                Exception e = new Exception("该文件不是WAVE文件");
                throw e;
            }*/

            System.arraycopy(wavBytes, p, riffSize, 0, 4); // 文件剩余长度
            p += 4;

            if (DataParseUtils.convertFoutUnsignLong(riffSize[3], riffSize[2], riffSize[1], riffSize[0]) != wavBytes.length - p)
            {
                //Exception e = new Exception("该WAVE文件损坏,文件长度与标记不一致");
                //throw e;
            }

            System.arraycopy(wavBytes, p, waveID, 0, 4);
            p += 4;

            if (DataParseUtils.convertFoutUnsignLong(waveID[3], waveID[2], waveID[1], waveID[0]) != 0x57415645)
            {
                Exception e = new Exception("该文件不是WAVE文件");
                throw e;
            }

            byte[] tmp = new byte[4];
            System.arraycopy(wavBytes, p, tmp, 0, 4);
            p += 4;

            if (DataParseUtils.convertFoutUnsignLong(tmp[3], tmp[2], tmp[1], tmp[0]) == 0x4A554E4B)
            {
                //包含junk标记的wav
                junkID = tmp;
                hasjunk = true;

                System.arraycopy(wavBytes, p, junklength, 0, 4);
                p += 4;

                long junklen = DataParseUtils.convertFoutUnsignLong(junklength[3], junklength[2], junklength[1], junklength[0]);


                //将不要的junk部分读出
                p += (int)junklen;

                //读fmt 标记
                System.arraycopy(wavBytes, p, fmtID, 0, 4);
                p += 4;
            }
            else if (DataParseUtils.convertFoutUnsignLong(tmp[3], tmp[2], tmp[1], tmp[0]) == 0x666D7420)
            {
                fmtID = tmp;
            }
            else
            {
                Exception e = new Exception("无法找到WAVE文件的junk和fmt标记");
                throw e;
            }


            if (DataParseUtils.convertFoutUnsignLong(fmtID[3], fmtID[2], fmtID[1], fmtID[0]) != 0x666D7420)
            {
                //fmt 标记
                Exception e = new Exception("无法找到WAVE文件fmt标记");
                throw e;
            }

            System.arraycopy(wavBytes, p, cksize, 0, 4);
            p += 4;

            long p_data_start = DataParseUtils.convertFoutUnsignLong(cksize[3], cksize[2], cksize[1], cksize[0]);
            int p_wav_start = (int)p_data_start + 8;
            byte[] tmp_waveType = new byte[2];
            System.arraycopy(wavBytes, p, tmp_waveType, 0, 2);
            p += 2;
            waveType = DataParseUtils.convertTwoUnsignInt(tmp_waveType[0], tmp_waveType[1]);

            if (waveType != 1)
            {
                // 非pcm格式,暂不支持
                Exception e = new Exception("WAVE文件不是pcm格式,暂时不支持");
                throw e;
            }

            //声道数
            System.arraycopy(wavBytes, p, channel, 0, 2);
            p += 2;

            //采样频率
            System.arraycopy(wavBytes, p, sample_rate, 0, 4);
            p += 4;

            int fs = (int)DataParseUtils.convertFoutUnsignLong(sample_rate[0], sample_rate[1], sample_rate[2], sample_rate[3]);

            //每秒钟字节数
            System.arraycopy(wavBytes, p, bytespersec, 0, 4);
            p += 4;

            //每次采样的字节大小,2为单声道,4为立体声道
            System.arraycopy(wavBytes, p, blocklen_sample, 0, 2);
            p += 2;

            //每个声道的采样精度,默认16bit
            System.arraycopy(wavBytes, p, bitNum, 0, 2);
            p += 2;

            System.arraycopy(wavBytes, p, tmp, 0, 2);
            p += 2;
            //寻找da标记
            while (DataParseUtils.convertTwoUnsignInt(tmp[1], tmp[0]) != 0x6461)
            {
                System.arraycopy(wavBytes, p, tmp, 0, 2);
                p += 2;
            }

            System.arraycopy(wavBytes, p, tmp, 0, 2);
            p += 2;

            if (DataParseUtils.convertTwoUnsignInt(tmp[1], tmp[0]) != 0x7461)
            {
                //ta标记
                Exception e = new Exception("无法找到WAVE文件data标记");
                throw e;
            }

            //wav数据byte长度
            byte[] data_size_byte = new byte[4];

            System.arraycopy(wavBytes, p, data_size_byte, 0, 4);
            p += 4;

            long DataSize = DataParseUtils.convertFoutUnsignLong(data_size_byte[0], data_size_byte[1], data_size_byte[2], data_size_byte[3]);
            //计算样本数
            long NumSamples = (long)DataSize / 2;

            if (NumSamples == 0)
            {
                NumSamples = (wavBytes.length - p) / 2;
            }

            short[] data = new short[(int) NumSamples];

            for (int i = 0; i < NumSamples; i++)
            {
                //读入2字节有符号整数
                byte[] tmp_sample = new byte[2];
                System.arraycopy(wavBytes, p, tmp_sample, 0, 2);
                p += 2;
                data[i] = (short)DataParseUtils.convertTwoUnsignInt(tmp_sample[0],tmp_sample[1]);
            }

            this.samples = data;
            this.sampleBytes = this.samplesToBytes(this.samples);
            this.sampleRate = fs;
            this.channels = DataParseUtils.convertTwoUnsignInt(channel[0],channel[1]);
            this.sampleWidth = DataParseUtils.convertTwoUnsignInt(bitNum[0],bitNum[1]) / 8;
            return true;
        }
        catch (Exception ex)
        {
            System.out.println(ex);
            return false;
        }
    }

    private byte[] serialize() {
        return null;
    }

    public byte[] getRawSamples() {
        return this.sampleBytes;
    }

    protected byte[] samplesToBytes(short[] samples){
        byte[] sampleBytes = new byte[samples.length * 2];
        for(int i = 0; i < samples.length; i++){
            byte[] sample = DataParseUtils.convertShortToBytes(samples[i], false);
            for(int j = 0; j < 2; j++){
                sampleBytes[2*i+j] = sample[j];
            }
        }
        return sampleBytes;
    }

    protected short[] bytesToSamples(byte[] sampleBytes){
        short[] data = new short[sampleBytes.length / 2];
        for (int i = 0; i < sampleBytes.length / 2; i++)
        {
            //读入2字节有符号整数
            byte[] tmp_sample = new byte[2];
            System.arraycopy(sampleBytes, 2 * i, tmp_sample, 0, 2);
            data[i] = (short)DataParseUtils.convertTwoUnsignInt(tmp_sample[0],tmp_sample[1]);
        }
        return data;
    }
}

class DataParseUtils:

class DataParseUtils {
    /**
     * 有符号,int 占 2 个字节
     */
    public static int convertTwoSignInt(byte b1, byte b2) { // signed
        return (b2 << 8) | (b1 & 0xFF);
    }

    /**
     * 有符号, int 占 4 个字节
     */
    public static int convertFourSignInt(byte b1, byte b2, byte b3, byte b4) {
        return (b4 << 24) | (b3 & 0xFF) << 16 | (b2 & 0xFF) << 8 | (b1 & 0xFF);
    }

    /**
     * 无符号,int 占 2 个字节
     */
    public static int convertTwoUnsignInt(byte b1, byte b2)      // unsigned
    {
        return (b2 & 0xFF) << 8 | (b1 & 0xFF);
    }

    /**
     * 无符号, int 占 4 个字节
     */
    public static long convertFoutUnsignLong(byte b1, byte b2, byte b3, byte b4) {
        return (long) (b4 & 0xFF) << 24 | (b3 & 0xFF) << 16 | (b2 & 0xFF) << 8 | (b1 & 0xFF);
    }

    public static byte[] convertShortToBytes(Short shortNumber, boolean big) {
        byte[] bytes = new byte[2];
        bytes[0] = (byte) (shortNumber & 0xff);
        bytes[1] = (byte) (shortNumber >> 8 & 0xff);
        if (big){
            byte tmp = bytes[0];
            bytes[0] = bytes[1];
            bytes[1] = tmp;
        }
        return bytes;
    }
}

class Common:

import java.io.FileInputStream;
import java.util.ArrayList;
import java.util.List;

public class Common {
    public static byte[] readBinFile(String filename) {
        FileInputStream input = null;
        try {
            List<Byte> byteList = new ArrayList();
            input = new FileInputStream(filename);
            byte[] buffer = new byte[1024];
            while (true) {
                int len = input.read(buffer);
                if (len == -1) {
                    break;
                }
                for(int i = 0; i < len; i++){
                    byteList.add(buffer[i]);
                }
            }
            byte[] byteArr = new byte[byteList.size()];
            for(int i = 0; i< byteArr.length; i++){
                byteArr[i] = byteList.get(i);
            }
            return byteArr;
        } catch (Exception e) {
            throw new RuntimeException(e);
        } finally {
            try {
                input.close();
            } catch (Exception e) {
                throw new RuntimeException(e);
            }
        }
    }
}

参考资料Refference

  1. AI柠檬,ASRT开源语音识别项目Java SDK. GitHub, https://github.com/nl8590687/ASRT_SDK_Java
版权声明
本博客的文章除特别说明外均为原创,本人版权所有。欢迎转载,转载请注明作者及来源链接,谢谢。
本文地址: https://blog.ailemon.net/2022/11/07/java-read-and-parse-wave-format-file/
All articles are under Attribution-NonCommercial-ShareAlike 4.0

关注“AI柠檬博客”微信公众号,及时获取你最需要的干货。


发表回复

您的电子邮箱地址不会被公开。 必填项已用 * 标注

4 × 2 =

如果您是第一次在本站发布评论,内容将在博主审核后显示,请耐心等待