0.6 Sept 20, 2012 Harry Liu, editorial updates
Sep 20, 2012 7 UCM-FM1288 V06
1.0 Overview
This document lists and describes all the control parameters accessible to the user for controlling the FM1288 Voice Processor.
The information contained in this document helps the user to ? Set up the FM-1288 hardware for different operation modes ; ? Debug the system data path during system integration ; and
? Adjust the various parameters to optimize for performance in different acoustic settings
2.0 Glossary and Notation
For a full list of terminology and glossary, please see References at the end of this document.
Throughout this document, these terminologies and abbreviations are used freely: Echo refers to the sound signal coupled from the loudspeaker output back into the microphone input via the acoustic path
Near end (NE) refers to the local conversation end point using the voice terminal equipped with the FM-1288
Far end (FE) refers to the conversation end point opposite to the near end
Double Talk(DT) refers to the situation during which Near end talker and Far end talker are talking at the same time
Receiving Direction refers to the line-in voice path going towards loudspeaker output Sending Direction refers to the microphone input voice path going towards line-out
Para[ address, ID ] is the notation used to denote a FM-1288 configurable parameter with name ID in alphanumeric and located at the data RAM address represented in hexadecimals
Sep 20, 2012 8 UCM-FM1288 V06
3.0 General configuration
The following is a description on how to set the general configuration by the host processor through the Serial Host Interface (SHI) port which is an I2C-compatible interface.
The default general configuration profile for FM-1288 is 1-microphone input mode, running under 8 kHz sampling rate for microphone in, line in and line out. On this voice processor, the Microphone input ADC and Speaker out DAC sampling rate are always fixed to be 16 KHz. The line in and line out signal pair could be configured to either 16 KHz or 8 KHz depending on the application configuration. The voice processing performed is always in the higher resolution of 16 KHz to ensure the best possible quality, and down-sampling to 8 KHz will be done automatically for proper rate matching when line in/out pair is configured to 8 KHz.
3.1 Default Profile Setup
1) Select internal profiles by setting Para[0x22F8, profile_index] as following: Para[0x22F8, profile_index] Internal profile 0x8000 1mic 8k (VDA setup)
2) Configure the user setting for the appliance.
3) To set Para[0x22FB, dv_parser_sync_flag] synchronization flag, the default value of it is 0x8000; writing it to 0 means that configuration has been completed. For example, VEC 0x22f8 0x0 : choose profile
0x2301 0x12 : user’s setting (Optional) 0x2302 0x1 : user’s setting (Optional)
0x22fb 0x0 : let IC run after configuration is done
3.2 Bandwidth Control Selection
Para[0x2301, sample_rate] Internal profile
0x0002 16KHz MIC/SPK sample rate, PCM/I2S data rate set to 16KHz.
0x0012 16KHz MIC/SPK sample rate, PCM/I2S data rate set to 8KHz
Sep 20, 2012 9 UCM-FM1288 V06
3.3 Single Microphone Hands-free Application Recommended Settings
For the Hands-free single-microphone mode in the most commonly deployed automotive applications, the following key voice processing components can be configured as follows: Table 1: Single Microphone Hands-Free Application Settings
Component
1 microphone hands free applicaton Recommended option
Mic-in HPF/ Line-in HPF On AEC On
BF (1-mic mode, no BF) Off BVE On DRC On SPK_ATTN On
PF On FENS On
Line-in AGC / Mic-in AGC Off Mic de-emphasis filter On
Line-in emphasis filter / de-emphasis filter Off Idle noise suppression On
Adjust noise suppression according to SNR On Noise residue stabilizer On Mic-in emphasis filter Off/On Spk-out HPF On BWE Off
Sep 20, 2012 10 UCM-FM1288 V06
4.0 Microphone Signal Conditions for FM-1288
Before starting any parameter tuning, the microphone mounting in the target system should be checked first to ensure that the system can work properly. Consideration of Air-tightness for microphones
Good mechanical design and mounting ensure air-tightness, which is very important to prevent performance degradation. To test airtight for main microphone, play the same NE (or FE) sound, then cover/uncover main microphone, and then record both signals respectively to find out the difference between the two. Then apply the same procedure to the reference microphone. A difference of 15dB or more is considered as a good condition. Table 2: Air-tight Condition for Microphones Mic sensitivity
without covered by mud (dB)
Mic sensitivity with covered by mud (dB) Criterion
NE Ane Bne Ane – Bne > 15db FE Afe Bfe Afe – Bfe > 15db
To verify if the microphone sensitivities have been properly set, first clear these: Para[0x3807, MAX_MIC0_readout], and Para[0x3809, MAX_LIN_readout].
Then play the typical test vectors to watch the 2 parameters which must not be saturated for microphone input and line-in signals.
The microphone signal should not be saturated except under usual conditions, such as wind blow, strong vibration, peak Far End talk volume, etc., especially during Far End talks. If on average the microphone is saturated more than once every several seconds, the
performance will be much degraded, and there is something wrong with the acoustics of the system.
Configure the system to play out a loud audio signal from the speaker, the peak digital value should not reach 0x7fff.
Confirm the delay between the echo and line-in signal for proper AEC operation: the delay should be larger than Para[0x2339, _tdaec_delay_length] and it ought to be a