Whole document tree

Chapter 4

Playing Back Audio

Playback is sometimes referred to as presentation or rendering. These are general terms that are applicable to other kinds of media besides sound. The essential feature is that a sequence of data is delivered somewhere for eventual perception by a user. If the data is time-based, as sound is, it must be delivered at the correct rate. With sound even more than video, it's important that the rate of data flow be maintained, because interruptions to sound playback often produce loud clicks or irritating distortion. The Java^TM Sound API is designed to help application programs play sounds smoothly and continuously, even very long sounds.

The previous chapter discussed how to obtain a line from the audio system or from a mixer. This chapter shows how to play sound through a line.

There are two kinds of line that you can use for playing sound: a Clip and a SourceDataLine. These two interfaces were introduced briefly under "The Line Interface Hierarchy" in Chapter 2, "Overview of the Sampled Package." The chief difference between the two is that with a Clip you specify all the sound data at one time, before playback, whereas with a SourceDataLine you keep writing new buffers of data continuously during playback. Although there are many situations in which you could use either a Clip or a SourceDataLine, the following criteria help identify which kind of line is better suited for a particular situation:

• Use a Clip when you have non-real-time sound data that can be preloaded into memory.

  For example, you might read a short sound file into a clip. If you want the sound to play back more than once, a Clip is more convenient than a SourceDataLine, especially if you want the playback to loop (cycle repeatedly through all or part of the sound). If you need to start the playback at an arbitrary position in the sound, the Clip interface provides a method to do that easily. Finally, playback from a Clip generally has less latency than buffered playback from a SourceDataLine. In other words, because the sound is preloaded into a clip, playback can start immediately instead of having to wait for the buffer to be filled.

• Use a SourceDataLine for streaming data, such as a long sound file that won't all fit in memory at once, or a sound whose data can't be known in advance of playback.

  As an example of the latter case, suppose you're monitoring sound input-that is, playing sound back as it's being captured. If you don't have a mixer that can send input audio right back out an output port, your application program will have to take the captured data and send it to an audio-output mixer. In this case, a SourceDataLine is more appropriate than a Clip. Another example of sound that can't be known in advance occurs when you synthesize or manipulate the sound data interactively in response to the user's input. For example, imagine a game that gives aural feedback by "morphing" from one sound to another as the user moves the mouse. The dynamic nature of the sound transformation requires the application program to update the sound data continuously during playback, instead of supplying it all before playback starts.

Using a Clip

You obtain a Clip as described earlier under "Getting a Line of a Desired Type" in Chapter 3, "Accessing Audio System Resources": Construct a DataLine.Info object with Clip.class for the first argument, and pass this DataLine.Info as an argument to the getLine method of AudioSystem or Mixer.

Setting Up the Clip for Playback

Obtaining a line just means you've gotten a way to refer to it; getLine doesn't actually reserve the line for you. Because a mixer might have a limited number of lines of the desired type available, it can happen that after you invoke getLine to obtain the clip, another application program jumps in and grabs the clip before you're ready to start playback. To actually use the clip, you need to reserve it for your program's exclusive use by invoking one of the following Clip methods:

   void open(AudioInputStream stream) 
   void open(AudioFormat format, byte[] data, int offset,
   int bufferSize)

After opening the clip, you can specify at what point in the data it should start playback, using Clip's setFramePosition or setMicroSecondPosition methods. Otherwise, it will start at the beginning. You can also configure the playback to cycle repeatedly, using the setLoopPoints method.

Starting and Stopping Playback

When you're ready to start playback, simply invoke the start method. To stop or pause the clip, invoke the stop method, and to resume playback, invoke start again. The clip remembers the media position where it stopped playback, so there's no need for explicit pause and resume methods. If you don't want it to resume where it left off, you can "rewind" the clip to the beginning (or to any other position, for that matter) using the frame- or microsecond-positioning methods mentioned above.

A Clip's volume level and activity status (active versus inactive) can be monitored by invoking the DataLine methods getLevel and isActive, respectively. An active Clip is one that is currently playing sound.

Using a SourceDataLine

Obtaining a SourceDataLine is similar to obtaining a Clip. See "Getting a Line of a Desired Type" in Chapter 3, "Accessing Audio System Resources."

Setting Up the SourceDataLine for Playback

Opening the SourceDataLine is also similar to opening a Clip, in that the purpose is once again to reserve the line. However, you use a different method, inherited from DataLine:

    void open(AudioFormat format) 

You can also stipulate a certain buffer length in bytes, using this variant:

    void open(AudioFormat format, int bufferSize) 

How would you select a buffer size? It depends on your program's needs.

To start with, shorter buffer sizes mean less latency. When you send new data, you hear it sooner. For some application programs, particularly highly interactive ones, this kind of responsiveness is important. For example, in a game, the onset of playback might need to be tightly synchronized with a visual event. Such programs might need a latency of less than 0.1 second. As another example, a conferencing application needs to avoid delays in both playback and capture. However, many application programs can afford a greater delay, up to a second or more, because it doesn't matter exactly when the sound starts playing, as long as the delay doesn't confuse or annoy the user. This might be the case for an application program that streams a large audio file using one-second buffers. The user probably won't care if playback takes a second to start, because the sound itself lasts so long and the experience isn't highly interactive.

On the other hand, shorter buffer sizes also mean a greater risk that you'll fail to write data fast enough into the buffer. If that happens, the audio data will contain discontinuities, which will probably be audible as clicks or breakups in the sound. Shorter buffer sizes also mean that your program has to work harder to keep the buffers filled, resulting in more intensive CPU usage. This can slow down the execution of other threads in your program, not to mention other programs.

So an optimal value for the buffer size is one that minimizes latency just to the degree that's acceptable for your application program, while keeping it large enough to reduce the risk of buffer underflow and to avoid unnecessary consumption of CPU resources. For a program like a conferencing application, delays are more annoying than low-fidelity sound, so a small buffer size is preferable. For streaming music, an initial delay is acceptable, but breakups in the sound are not, so a larger buffer size-say, a second-is preferable. (Note that high sample rates make the buffers larger in terms of the number of bytes, which are the units for measuring buffer size in the DataLine API.)

Instead of using the open method described above, it's also possible to open a SourceDataLine using Line's open() method, without arguments. In this case, the line is opened with its default audio format and buffer size. However, you can't change these later. If you want to know the line's default audio format and buffer size, you can invoke DataLine's getFormat and getBufferSize methods, even before the line has ever been opened.

Starting and Stopping Playback

Once the SourceDataLine is open, you can start playing sound. You do this by invoking DataLine's start method, and then writing data repeatedly to the line's playback buffer.

The start method permits the line to begin playing sound as soon as there's any data in its buffer. You place data in the buffer by the following method:

    int write(byte[] b, int offset, int length) 

The line begins sending data as soon as possible to its mixer. When the mixer itself delivers the data to its target, the SourceDataLine generates a START event. (In a typical implementation of the Java Sound API, the delay between the moment that the source line delivers data to the mixer and the moment that the mixer delivers the data to its target is negligible-that is, much less than the time of one sample.) This START event gets sent to the line's listeners, as explained below under "Monitoring a Line's Status." The line is now considered active, so the isActive method of DataLine will return true. Notice that all this happens only once the buffer contains data to play, not necessarily right when the start method is invoked. If you invoked start on a new SourceDataLine but never wrote data to the buffer, the line would never be active and a START event would never be sent. (However, in this case, the isRunning method of DataLine would return true.)

So how do you know how much data to write to the buffer, and when to send the second batch of data? Fortunately, you don't need to time the second invocation of write to synchronize with the end of the first buffer! Instead, you can take advantage of the write method's blocking behavior:

• The method returns as soon as the data has been written to the buffer. It doesn't wait until all the data in the buffer has finished playing. (If it did, you might not have time to write the next buffer without creating a discontinuity in the audio.)
• It's all right to try to write more data than the buffer will hold. In this case, the method blocks (doesn't return) until all the data you requested has actually been placed in the buffer. In other words, one buffer's worth of your data at a time will be written to the buffer and played, until the remaining data all fits in the buffer, at which point the method returns. Whether or not the method blocks, it returns as soon as the last buffer's worth of data from this invocation can be written. Again, this means that your code will in all likelihood regain control before playback of the last buffer's worth of data has finished.
• While in many contexts it is fine to write more data than the buffer will hold, if you want to be certain that the next write issued does not block, you can limit the number of bytes you write to the number that DataLine's available method returns.

// read chunks from a stream and write them to a source data 
line 
line.start();
while (total < totalToRead && !stopped)}
    numBytesRead = stream.read(myData, 0, numBytesToRead);
    if (numBytesRead == -1) break;
    total += numBytesRead; 
    line.write(myData, 0, numBytesRead);

}

Since write returns before all the data has finished playing, how do you learn when the playback has actually completed? One way is to invoke the drain method of DataLine after writing the last buffer's worth of data. This method blocks until all the data has been played. When control returns to your program, you can free up the line, if desired, without fear of prematurely cutting off the playback of any audio samples:

line.write(b, offset, numBytesToWrite); 
//this is the final invocation of write
line.drain();
line.stop();
line.close();
line = null;

A SourceDataLine generates a STOP event whenever the flow of data has been stopped, whether this stoppage was initiated by the drain method, the stop method, or the flush method, or because the end of a playback buffer was reached before the application program invoked write again to provide new data. A STOP event doesn't necessarily mean that the stop method was invoked, and it doesn't necessarily mean that a subsequent invocation of isRunning will return false. It does, however, mean that isActive will return false. (When the start method has been invoked, the isRunning method will return true, even if a STOP event is generated, and it will begin to return false only once the stop method is invoked.) It's important to realize that START and STOP events correspond to isActive, not to isRunning.

Monitoring a Line's Status

Once you have started a sound playing, how do you find when it's finished? We saw one solution above-invoking the drain method after writing the last buffer of data-but that approach is applicable only to a SourceDataLine. Another approach, which works for both SourceDataLines and Clips, is to register to receive notifications from the line whenever the line changes its state. These notifications are generated in the form of LineEvent objects, of which there are four types: OPEN, CLOSE, START, and STOP.

Any object in your program that implements the LineListener interface can register to receive such notifications. To implement the LineListener interface, the object simply needs an update method that takes a LineEvent argument. To register this object as one of the line's listeners, you invoke the following Line method:

    public void addLineListener(LineListener listener)

Along the same lines, if you need to know when the line is opened, closed, or started, you use the same mechanism. LineEvents are generated by different kinds of lines, not just Clips and SourceDataLines. However, in the case of a Port you can't count on getting an event to learn about a line's open or closed state. For example, a Port might be initially open when it's created, so you don't invoke the open method and the Port doesn't ever generate an OPEN event. (See "Selecting Input and Output Ports" in Chapter 3, "Accessing Audio System Resources.")

Synchronizing Playback on Multiple Lines

If you're playing back multiple tracks of audio simultaneously, you probably want to have them all start and stop at exactly the same time. Some mixers facilitate this behavior with their synchronize method, which lets you apply operations such as open, close, start, and stop to a group of data lines using a single command, instead of having to control each line individually. Furthermore, the degree of accuracy with which operations are applied to the lines is controllable.

To find out whether a particular mixer offers this feature for a specified group of data lines, invoke the Mixer interface's isSynchronizationSupported method:

boolean isSynchronizationSupported(Line[] lines, 
  boolean  maintainSync)

Processing the Outgoing Audio

Some source data lines have signal-processing controls, such as gain, pan, reverb, and sample-rate controls. Similar controls, especially gain controls, might be present on the output ports as well. For more information on how to determine whether a line has such controls, and how to use them if it does, see Chapter 6, "Processing Audio with Controls."