使用 Java 合并 2 个 AAC 文件
Posted
技术标签:
【中文标题】使用 Java 合并 2 个 AAC 文件【英文标题】：Combine 2 AAC files using Java 【发布时间】：2012-09-25 00:54:04 【问题描述】：我试图将 2 个 AAC 文件合并为一个，我发现在 AAC 文件中 - 标题元素一直存在到 FF8 字节，然后接下来的 4 个字节包含 AAC 数据的长度。我试着维护1个header数组，把2个AAC文件的大小相加，然后把两个文件的数据缓冲区一个接一个地加进去。
生成的文件仅播放第一个 AAC 文件。这是代码sn-p。
FileInputStream fs = new FileInputStream("./res/after.aac");
dis = new DataInputStream(fs);
headerData = new byte[0xFF8];
dis.read(headerData);
int lengthTotal = dis.readInt();
System.out.println("Length of After == "+lengthTotal);
dis.readInt();
data = new byte[dis.available()];
dis.readFully(data);
dis.close();
dis = null;
fs.close();
fs = null;
fs = new FileInputStream("./res/continue.aac");
dis = new DataInputStream(fs);
dis.skipBytes(0xFF8);
int length = dis.readInt();
System.out.println("Length of Ahead == "+length);
lengthTotal = lengthTotal + length -8;
System.out.println("Total Length== "+lengthTotal);
dis.readInt();
newData = new byte[dis.available()];
dis.read(newData);
FileOutputStream fos = new FileOutputStream("./res/combine.aac");
DataOutputStream dos = new DataOutputStream(fos);
dos.write(headerData);
dos.writeInt(lengthTotal);
dos.writeBytes("mdat");
dos.write(data);
dos.write(newData);
我知道在第 56 个字节中有关于 AAC 文件持续时间的信息，但我无法弄清楚。有人可以帮我吗？
【问题讨论】：
我也在尝试同样的方法...你解决了吗？有什么解决办法吗？ 【参考方案1】：我建议你看看jaad库中aac文件是如何解析的，特别有趣的是ADTS处理代码here和ADIF头解析here。
【讨论】：
【参考方案2】：好吧，我不能告诉你你做错了什么。但我可以告诉你如何做你想做的事。
首先创建一个General Helper Functions：
public static class General
public static void CopyBytes(byte[] dst, int dstOffset, byte[] src)
Buffer.BlockCopy(src, 0, dst, dstOffset, src.Length);
public static class BitHelper
public static int Read(ref ulong x, int length)
int r = (int)(x >> (64 - length));
x <<= length;
return r;
public static int Read(byte[] bytes, ref int offset, int length)
int startByte = offset / 8;
int endByte = (offset + length - 1) / 8;
int skipBits = offset % 8;
ulong bits = 0;
for (int i = 0; i <= Math.Min(endByte - startByte, 7); i++)
bits |= (ulong)bytes[startByte + i] << (56 - (i * 8));
if (skipBits != 0) Read(ref bits, skipBits);
offset += length;
return Read(ref bits, length);
public static void Write(ref ulong x, int length, int value)
ulong mask = 0xFFFFFFFFFFFFFFFF >> (64 - length);
x = (x << length) | ((ulong)value & mask);
public static byte[] CopyBlock(byte[] bytes, int offset, int length)
int startByte = offset / 8;
int endByte = (offset + length - 1) / 8;
int shiftA = offset % 8;
int shiftB = 8 - shiftA;
byte[] dst = new byte[(length + 7) / 8];
if (shiftA == 0)
Buffer.BlockCopy(bytes, startByte, dst, 0, dst.Length);
else
int i;
for (i = 0; i < endByte - startByte; i++)
dst[i] = (byte)((bytes[startByte + i] << shiftA) | (bytes[startByte + i + 1] >> shiftB));
if (i < dst.Length)
dst[i] = (byte)(bytes[startByte + i] << shiftA);
dst[dst.Length - 1] &= (byte)(0xFF << ((dst.Length * 8) - length));
return dst;
public static class BitConverterBE
public static ulong ToUInt64(byte[] value, int startIndex)
return
((ulong)value[startIndex ] << 56) |
((ulong)value[startIndex + 1] << 48) |
((ulong)value[startIndex + 2] << 40) |
((ulong)value[startIndex + 3] << 32) |
((ulong)value[startIndex + 4] << 24) |
((ulong)value[startIndex + 5] << 16) |
((ulong)value[startIndex + 6] << 8) |
((ulong)value[startIndex + 7] );
public static uint ToUInt32(byte[] value, int startIndex)
return
((uint)value[startIndex ] << 24) |
((uint)value[startIndex + 1] << 16) |
((uint)value[startIndex + 2] << 8) |
((uint)value[startIndex + 3] );
public static ushort ToUInt16(byte[] value, int startIndex)
return (ushort)(
(value[startIndex ] << 8) |
(value[startIndex + 1] ));
public static byte[] GetBytes(ulong value)
byte[] buff = new byte[8];
buff[0] = (byte)(value >> 56);
buff[1] = (byte)(value >> 48);
buff[2] = (byte)(value >> 40);
buff[3] = (byte)(value >> 32);
buff[4] = (byte)(value >> 24);
buff[5] = (byte)(value >> 16);
buff[6] = (byte)(value >> 8);
buff[7] = (byte)(value );
return buff;
public static byte[] GetBytes(uint value)
byte[] buff = new byte[4];
buff[0] = (byte)(value >> 24);
buff[1] = (byte)(value >> 16);
buff[2] = (byte)(value >> 8);
buff[3] = (byte)(value );
return buff;
public static byte[] GetBytes(ushort value)
byte[] buff = new byte[2];
buff[0] = (byte)(value >> 8);
buff[1] = (byte)(value );
return buff;
public static class BitConverterLE
public static byte[] GetBytes(ulong value)
byte[] buff = new byte[8];
buff[0] = (byte)(value );
buff[1] = (byte)(value >> 8);
buff[2] = (byte)(value >> 16);
buff[3] = (byte)(value >> 24);
buff[4] = (byte)(value >> 32);
buff[5] = (byte)(value >> 40);
buff[6] = (byte)(value >> 48);
buff[7] = (byte)(value >> 56);
return buff;
public static byte[] GetBytes(uint value)
byte[] buff = new byte[4];
buff[0] = (byte)(value );
buff[1] = (byte)(value >> 8);
buff[2] = (byte)(value >> 16);
buff[3] = (byte)(value >> 24);
return buff;
public static byte[] GetBytes(ushort value)
byte[] buff = new byte[2];
buff[0] = (byte)(value );
buff[1] = (byte)(value >> 8);
return buff;
现在实现 Audio Helper 类和接口：
interface IAudioWriter
void WriteChunk(byte[] chunk, uint timeStamp);
void Finish();
string Path get;
class AACWriter : IAudioWriter
string _path;
FileStream _fs;
int _aacProfile;
int _sampleRateIndex;
int _channelConfig;
public AACWriter(string path)
_path = path;
_fs = new FileStream(path, FileMode.Create, FileAccess.Write, FileShare.Read, 65536);
public void WriteChunk(byte[] chunk, uint timeStamp)
if (chunk.Length < 1) return;
if (chunk[0] == 0) // Header
if (chunk.Length < 3) return;
ulong bits = (ulong)BitConverterBE.ToUInt16(chunk, 1) << 48;
_aacProfile = BitHelper.Read(ref bits, 5) - 1;
_sampleRateIndex = BitHelper.Read(ref bits, 4);
_channelConfig = BitHelper.Read(ref bits, 4);
if ((_aacProfile < 0) || (_aacProfile > 3))
throw new Exception("Unsupported AAC profile.");
if (_sampleRateIndex > 12)
throw new Exception("Invalid AAC sample rate index.");
if (_channelConfig > 6)
throw new Exception("Invalid AAC channel configuration.");
else // Audio data
int dataSize = chunk.Length - 1;
ulong bits = 0;
// Reference: WriteADTSHeader from FAAC's bitstream.c
BitHelper.Write(ref bits, 12, 0xFFF);
BitHelper.Write(ref bits, 1, 0);
BitHelper.Write(ref bits, 2, 0);
BitHelper.Write(ref bits, 1, 1);
BitHelper.Write(ref bits, 2, _aacProfile);
BitHelper.Write(ref bits, 4, _sampleRateIndex);
BitHelper.Write(ref bits, 1, 0);
BitHelper.Write(ref bits, 3, _channelConfig);
BitHelper.Write(ref bits, 1, 0);
BitHelper.Write(ref bits, 1, 0);
BitHelper.Write(ref bits, 1, 0);
BitHelper.Write(ref bits, 1, 0);
BitHelper.Write(ref bits, 13, 7 + dataSize);
BitHelper.Write(ref bits, 11, 0x7FF);
BitHelper.Write(ref bits, 2, 0);
_fs.Write(BitConverterBE.GetBytes(bits), 1, 7);
_fs.Write(chunk, 1, dataSize);
public void Finish()
_fs.Close();
public string Path
get
return _path;
现在你需要自己做的是，从第一个 AAC 文件中逐个读取块并写入它们，然后从第二个 AAC 文件中逐个读取块并将它们附加到中间文件。
注意，上面的代码是 C#，所以你必须使用 wrapping 来模拟 C# 的 ref 效果，只需替换：
ref Type variable_name
与：
_<Type> variable_name
【讨论】：
即使这样仍然不够，因为有一堆 MPEG4 元数据（如stsz 样本大小表）必须智能组合。
您的建议属于 MP4 容器的范围。每个块中的ADTSHeader 为解码器提供了足够的信息以正确解码音频，即 AAC 原始文件。 @duskwuff
示例代码中对"mdat"的引用让我觉得它实际上是一个扩展名错误的MPEG4。以上是关于使用 Java 合并 2 个 AAC 文件的主要内容，如果未能解决你的问题，请参考以下文章