ST STM32F103 series Installation and operating instructions

April 2009 Doc ID 15349 Rev 2 1/15
AN2931
Application note
Implementing the ADPCM algorithm
in high-density STM32F103xx microcontrollers
Introduction
This application note describes the ADPCM audio firmware codec, and provides a
demonstration firmware showing how to play ADPCM files using the high-density
STM32F103xx I2S feature with an external DAC.
This application note is based on “AN2739: How to use the high-density STM32F103xx
microcontroller to play audio files with an external I2S audio codec”. The features linked to
the SPI, I2S and the external DAC will be skipped here because they are already explained
in AN2739. It is therefore highly recommended to refer to AN2739 to fully understand this
application note.
www.st.com

Contents AN2931
2/15 Doc ID 15349 Rev 2
Contents
1 ADPCM algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.1 General overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2 ADPCM algorithm implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 ADPCM algorithm functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3.1 ADPCM_Encode function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3.2 ADPCM_Decode function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 Implementation example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2 ADPCM file generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3 ADPCM file loading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.4 ADPCM file playing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
4 Revision history . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

AN2931 List of tables
Doc ID 15349 Rev 2 3/15
List of tables
Table 1. ADPCM algorithm functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Table 2. ADPCM_Encode function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Table 3. ADPCM_Decode function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Table 4. Document revision history . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

List of figures AN2931
4/15 Doc ID 15349 Rev 2
List of figures
Figure 1. ADPCM file generator. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Figure 2. DFU File Manager . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Figure 3. DFU File Manager source browser . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Figure 4. DfuSe software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

AN2931 ADPCM algorithm
Doc ID 15349 Rev 2 5/15
1 ADPCM algorithm
1.1 General overview
Adaptive differential pulse-code modulation, or simply ADPCM, is an audio algorithm for
waveform coding, which consists in predicting the current signal value from previous values,
and transmitting only the difference between the real and the predicted value.
In plain pulse-code modulation (PCM), the real or actual signal value is transmitted.
The advantage of ADPCM is that the difference between the predicted signal value and the
actual signal value is usually quite small, which means it can be represented using fewer
bits than the corresponding PCM value.
Depending on the desired quality and compression ratio, the differential signal is quantized
using 4 (2 bit), 8 (3 bit), 16 (4 bit) or 32 (5 bit) levels.
Many implementations of the ADPCM algorithm exist. They differ by the quantization and
the prediction patterns.
In this application note, we provide a 4-bit quantization algorithm, developed by the
Interactive Multimedia Association (IMA), IMA ADPCM.
IMA ADPCM was chosen for several reasons:
●it can operate at different sampling rates between 8 kHz and 44.10 kHz
●it guarantees good quality with low CPU usage and memory size requirements
●it has widespread implementations, for instance in the Windows and MAC operating
systems
The IMA ADPCM algorithm is fully described in a document published by the IMA Digital
Audio Focus and Technical Working Groups: “Recommended Practices for Enhancing
Digital Audio Compatibility in Multimedia Systems” revision 3.
1.2 ADPCM algorithm implementation
The IMA ADPCM algorithm provided in this application note is used to encode audio files
that have the following specification:
●Audio format: PCM
●Audio sample size: 16 bits
●Channels: 1 (mono)
●Audio sample rate: from 8 kHz to 44.1 kHz
Each 16-bit PCM sample is encoded into a 4-bit ADPCM sample, which gives a
compression rate equal to ¼.
The implementation of the IMA ADPCM algorithm consists in two functions, one that codes
and the other that decodes the audio samples.
The ADPCM firmware is composed of two files:
a) adpcm.c: it contains the source code of the two ADPCM functions that perform the
coding and decoding.
b) adpcm.h: it is the header file of adpmc.c. It should be included in the files where
the ADPCM functions are called.

ADPCM algorithm AN2931
6/15 Doc ID 15349 Rev 2
1.3 ADPCM algorithm functions
Table 1 describes the ADPCM functions.
1.3.1 ADPCM_Encode function
Table 2 describes the ADPCM_Encode function.
Table 2. ADPCM_Encode function
The ADPCM_Encode function returns a byte containing the 4-bit ADPCM sample. The
software has to store every two ADPCM samples into one byte to save memory space.
Example:
//Input: pcm_sample1 and pcm_sample2; two 16-bit PCM samples.
//Output: adpcm_byte; two 4-bit ADPCM samples stored into one byte.
uint8_t code;
/* Encode the first 16-bit sample */
code = ADPCM_Encode(pcm_sample1);
/* Store the first 4-bit sample */
adpcm_byte = code;
/* Encode the second 16-bit sample */
code = ADPCM_Encode(pcm_sample2);
/* Store the second 4-bit sample */
adpcm_byte |= (code << 4);/* The adpcm_byte contains two 4-bit ADPCM
samples */
1.3.2 ADPCM_Decode function
Table 3 describes the ADPCM_Decode function.
Table 1. ADPCM algorithm functions
Function name Description
ADPCM_Encode Encodes a 16-bit PCM sample into a 4-bit ADPCM sample.
ADPCM_Decode Decodes a 4-bit ADPCM sample into a 16-bit PCM sample.
Function name ADPCM_Encode
Prototype uint8_t ADPCM_Encode(int16_t sample);
Behavior description Encodes a 16-bit PCM sample into 4-bit ADPCM sample
Input parameter Sample: a 16-bit PCM sample
Output parameter None
Return parameter The encoded ADPCM 4-bit sample loaded into a byte

AN2931 ADPCM algorithm
Doc ID 15349 Rev 2 7/15
Table 3. ADPCM_Decode function
The input of the ADPCM_Decode function is a byte that contains the 4-bit ADPCM sample.
The software has to extract the 4-bit ADPCM data and store them into a byte before calling
the ADPCM_Decode function.
Example:
//Input: adpcm_byte; two 4-bit ADPCM samples stored into one byte.
//Output: pcm_sample1 and pcm_sample2; two 16-bit PCM samples.
uint8_t code;
/* Extract the first ADPCM 4-bit sample */
code = (adpcm_byte & 0x0F);
/* Decode the first ADPCM sample */
pcm_sample1 = ADPCM_Decode(code);
/* Extract the second ADPCM sample */
code = (adpcm_byte >> 4);
/* Decode the second ADPCM sample */
pcm_sample2 = ADPCM_Decode(code);
Function name ADPCM_Decode
Prototype int16_t ADPCM_Decode(uint8_t code);
Behavior description Decodes a 4-bit ADPCM sample into a 16-bit PCM sample
Input parameter Code: an 8-bit data item whose 4 LSBs contain the encoded ADPCM
sample
Output parameter None
Return parameter A 16-bit PCM sample

Implementation example AN2931
8/15 Doc ID 15349 Rev 2
2 Implementation example
2.1 Description
You can run the provided example on the STM3210E-EVAL board. It is a typical audio
application that consists in:
●first encoding a PCM file into the ADPCM format using a PC software (This software is
called muse.exe, and was developed by ST. It implements the same ADPCM algorithm
encoder as the one that runs on STM32F103xx performance line devices. muse.exe is
included in the zipped software package that comes with this application note: please
refer to Section 2.2 for how to use this tool.)
●then loading the encoded file into a Flash memory
●and finally decoding the file and playing it using the high-density STM32F103xx
microcontroller.
On the STM3210E-EVAL board, the 128 Mbit NOR Flash memory stores the ADPCM file
and the firmware decoder that runs on STM32F103xx devices. It then decodes the stream
and, with the I2S external DAC, plays back the audio samples.
The hardware environment is based on the application note “AN2739: How to use the high-
density STM32F103xx microcontroller to play audio files with an external I2S audio codec”,
so please refer to that application note for more information.
To run this demo, three steps are required:
1. generating the ADPCM file
2. loading the ADPCM file into the NOR Flash memory
3. playing the ADPCM file
2.2 ADPCM file generation
The application note package contains a PC software called Muse to encode PCM files into
the ADPCM format.
Figure 1. ADPCM file generator

AN2931 Implementation example
Doc ID 15349 Rev 2 9/15
If you run the Muse.exe program and press the Source File Browse button, the File
Selection dialog appears, prompting you to select a WAV source file. As mentioned before,
the WAV file should meet the following requirements:
●Audio format: PCM
●Audio sample size: 16 bits
●Channels: 1 (mono)
●Audio sample rate: from 8 kHz to 44.1 kHz
If your WAV file does not satisfy these conditions, you must previously convert it using an
audio converter tool, like sndrec32.exe, which is the Microsoft®default program for audio
records.
Once you have selected the file to be converted, the New File Selection dialog appears,
prompting you to select the save location of your ADPCM converted file.
You can retrieve the PCM format of the encoded audio file by ticking the Make a recovered
PCM file check box. In this case, the generated ADPCM file is decoded and formatted into a
WAV file. This WAV file is called Recover.wav and it is generated in the same directory as
your WAV source file. In this way you can listen to both the original WAV file and the
recovered one to compare the sound quality.
2.3 ADPCM file loading
Use the DfuSe software to load the ADPCM file into the NOR Flash memory. To do so, two
steps are necessary:
1. generating the .dfu image
2. loading the .dfu image into the NOR Flash memory
Generating the .dfu image
To generate the .dfu image, you have to use the DFU File Manager software provided within
the DfuSe package. For that, open the DFU File Manager program and choose the DFU file
generation option. The window shown in Figure 2 opens. You have to select 2 for the target
ID, meaning that the file is loaded into the NOR Flash memory. Then, click on the Multi
BIN... button to open the source browser and select your ADPCM file.

Implementation example AN2931
10/15 Doc ID 15349 Rev 2
Figure 2. DFU File Manager
The NOR Flash memory is mapped on the FSMC interface from address 0x6400 0000,
which is the start address of the ADPCM file storage area. So, you have to set the address
of the .dfu image to 0x6400 0000 (see Figure 3). Finally, click on Add to list and then
Generate to finish the DFU file generation.
Figure 3. DFU File Manager source browser
Loading the DFU image
To load the DFU image into the NOR Flash memory you have to:
1. connect the STM3210E-EVAL board to a PC using a USB cable
2. load and run the device firmware upgrade (DFU) demonstration on the STM3210E-
EVAL board. You can find this firmware in the STM3210xxx USB development kit
3. start the DfuSe software on the PC (see Figure 4)