4.3 Data Transfer Methods

(Page 4 of 8 in this chapter)

AG Access employs a double-buffering scheme when recording and playing speech files. This means that when the AG board finishes processing a speech buffer, the subsequent speech buffer must already be allocated and submitted to AG Access by the application.
Failure to maintain pace with the AG board causes underruns in the record or playback. Underruns are manifested as gaps in the speech. AG Access counts the number of underruns that occur but not their duration. The underrun count is retrieved via adiGetRecordStatus and adiGetPlayStatus functions.
Note: It is important not to submit small buffers (e.g., buffers that hold less than one second of data) because small buffers can also cause underruns. The data throughput for a given encoding method can be derived from the values returned by adiGetEncodingInfo.
AG Access provides three methods by which the application can transfer speech data to and from the AG board: Single Memory Transaction: The application submits a single data buffer to AG Access. Asynchronous Transfer: The application serially submits multiple data buffers by exchanging commands and events with AG Access. Callback Transfer: AG Access manages the buffers and invokes an application callback function to retrieve or store data.
The functions used to initiate play or record depends upon the data transfer method selected as shown below:

Single Memory

Asynchronous

Callback

Play

adiPlayFromMemory

adiPlayAsync

adiStartPlaying

Record

adiRecordToMemory

adiRecordAsync

adiStartRecording
The following three sections describe the general mechanics for the three transfer methods. These methods are described in greater detail in Section 4.5 and Section 4.6. The AG Access development kit contains the playrec demo, which demonstrates the play and record functions for both the callback and asynchronous methods.

	Single Memory	Asynchronous	Callback
Play	adiPlayFromMemory	adiPlayAsync	adiStartPlaying
Record	adiRecordToMemory	adiRecordAsync	adiStartRecording

4.3.1 Single Memory Transaction

If the application invokes adiPlayFromMemory or adiRecordToMemory, it supplies a single buffer that is retained by AG Access for the duration of the function. AG Access divides the application buffer into AG physical segments and performs all handshaking with the board.
A play buffer can be shared by multiple instances of the play function (within the same process) but the record buffer must be unique for each active recording instance. When AG Access delivers the ADIEVN_PLAY_DONE or ADIEVN_RECORD_DONE event to the application, the buffer is then available for reuse or disposal. Advantage: Simple -- minimizes application interaction. Disadvantage: Consumes a large virtual address space for large speech files. May experience latency reading or writing large files to and from storage.

4.3.2 Asynchronous Transfer

The asynchronous transfer method gives you maximum latitude with buffer address, size, and submission. An initial buffer is submitted when the play/record function is started via adiPlayAsync or adiRecordAsync. AG Access generates events whenever more speech data is required (play) or when a buffer is filled (record). The application is responsible for submitting buffers in a timely manner via adiSubmitPlayBuffer or adiSubmitRecordBuffer.

Advantage:

Programmer-controlled buffer addresses and sizes.
Allows asynchronous access to storage medium.

Disadvantage:

More complicated to program.

4.3.3 Callback Transfer

The callback transfer method strikes a balance between simplicity and resource consumption. AG Access allocates the buffers and invokes an application callback function whenever a buffer needs to be filled (playing) or when a buffer needs to be emptied (record). Within the callback, the application synchronously accesses the storage medium before returning. Advantage: Minimizes virtual memory consumption. Simple to implement. Disadvantage: Application cannot control the buffer addresses or sizes. Requires synchronous access to storage medium.

(Page 4 of 8 in this chapter)

Advantage:	Simple -- minimizes application interaction.
Disadvantage:	Consumes a large virtual address space for large speech files. May experience latency reading or writing large files to and from storage.

Advantage:	Programmer-controlled buffer addresses and sizes. Allows asynchronous access to storage medium.
Disadvantage:	More complicated to program.

Advantage:	Minimizes virtual memory consumption. Simple to implement.
Disadvantage:	Application cannot control the buffer addresses or sizes. Requires synchronous access to storage medium.