QNX Technical Articles

Date of this edition: July 16, 2013

Target OS: This software is compatible with target systems running QNX^® Neutrino^® 6.3.2 through 6.5.0.

Host OS: You must have already installed the QNX^® Software Development Platform:

• on a self-hosted QNX Neutrino development system on a physical machine, or as a virtual machine on VMware Workstation 6.5 or 7.0, VMware Player 3.0, or Microsoft VirtualPC 2007
• or on one of the following development hosts:

  Version	Microsoft Windows	Linux
  QNX Momentics 6.3.2	Windows Vista, XP SP2, or 2000 SP4	Linux Red Hat Enterprise Workstation 4 or 5, Red Hat Fedora Core 6 or 7, Ubuntu 6.0.6 LTS or 7.0.4, or SUSE 10
  QNX SDP 6.4.0	Windows Vista, Vista 64-bit, XP SP2 or SP3, or 2000 SP4	Linux Red Hat Enterprise Workstation 4 or 5, Red Hat Enterprise Server 5.1 64-bit, Red Hat Fedora Core 6 or 7, Ubuntu 6.0.6 LTS or 7, or SUSE 10
  QNX SDP 6.4.1	Windows Vista, Vista 64-bit, XP SP2 or SP3, or 2000 SP4	Linux Red Hat Enterprise Workstation 4.0 or 5.0, Red Hat Enterprise Server 5.1 64-bit, Red Hat Fedora 10, Ubuntu 8.04 LTS or 8.10, or SUSE 11
  QNX SDP 6.5.0	Windows 7 Professional 32- and 64-bit, Vista Business 32- and 64-bit, XP Professional SP3, or 2000 SP4	Linux Red Hat Enterprise Linux 5.4 Desktop 32- and 64-bit, Red Hat Fedora 12, Ubuntu Workstation 9.10 32- and 64-bit, or openSUSE 11.2

Contents

• What's in this package?
  • New acoustic processing options
  • API changes
  • Binaries
  • Sample source code
• Fixed issues
• Known issues
• Benchmarks
  • ROM benchmarks
  • CPU Load/RAM Usage benchmarks
  • Processing Time
• Documentation
• Technical support

Throughout this document, you may see reference numbers associated with particular issues, changes, etc. When corresponding with our Technical Support staff about a given issue, please quote the relevant reference number. You might also find the reference numbers useful for tracking issues as they become fixed.

For the most up-to-date version of these release notes, go to our website, www.qnx.com, log into your myQNX account, and then go to the Download Center.

What's in this package?

New acoustic processing options

The core features of QNX Aviage Acoustic Processing previous version 1.3 include:

• Binary configuration file load support
• Wide-band support (WBND)
• Send-side(Mic-In audio signal)
  • Acoustic echo cancellation (AEC)
  • Noise reduction (NR)
  • Automatic gain control (AGC)
  • Parametric equalization (PEQ)
  • Dual-channel capability (MIX)
  • High frequency encoding (HFE)
  • Wind buffet suppression (WNDB)
  • Dynamic limiter (SDL)
  • Dynamic parametric equalization (DPEQ)
• Receive-side(Recv-In audio signal)
  • Bandwidth extension (RBWE)
  • Parametric equalization (RPEQ)
  • Automatic gain control (RAGC)
  • Dynamic limiter (RDL)
  • Dynamic level control (RDLC)
  • Electrical noise suppression (RENS)
  • Double talk filter (RDTF)
• Diagnostics API (DIAG)
  • Output clipping statistics
• Remote control server (RCS)
• Windows remote control client (QWALive)
  • Binary configuration file save support
  • Audio signal injection support
  • Coherence measurements
  • Total harmonic distortion measurements

The new features in the QNX AAP 2.0 include:

• Send-side(Mic-In audio signal)
  • Automatic delay compensation (ADC).
  • Off-axis rejection (OAR).
  • Dynamic noise removal (DNR).
  • Low-frequency reconstruction (LFR).
  • Nonlinear distortion half duplex threshold (NLD)
• Remote control server (RCS)
  • Injection error reporting.
  • Serial port communication option.
  • New API for post-initialization configuration.
• Windows QWALive remote control client
  • Ability to connect to a target through serial port.
  • Ability to set a name for each target connection.
  • Ability to set a custom color for each target connection.
  • Ability to run coherence and distortion measurements for the second microphone channel.
  • Ability to monitor and listen to an incoming audio stream.
  • Ability to pause streaming.
  • Ability to save binary configuration files in text format.
  • New audio path test in printable and HTML viewable format.
  • New injection dropouts error indicator.
  • New "Select Baseline Config" button in the Binary Config tab to select the minimum subset of parameters recommended for binary configuration based on the features (switches) that are turned on.
  • Existing EQ presets are now displayed in the "SavePEQ Preset" form.
  • New button "Delete Preset" added in the "LoadPEQ Preset" form.
  • New "Adjustment(Post Init)" tab in the Binary Config for post-initialization adjustment of parameters.
  • Parameter qwaAECFiltLen can now be changed using the "AEC Filter Length" control on the reinitialize page if the Enable Large Room Support check box is selected in "App Options" tab.
  • Parameter qwaAECResEchoRT60ms can now be changed using the "Reverberation Time" control on the Send page if the Enable Large Room Support check box is selected in "App Options" tab.

To support the new features the following parameters were added:

• qwaAutoDelayCompSwitch
• qwaMixOARSwitch
• qwaLFRSwitch
• qwaLFINRMaxFreq
• qwaMixOARClosestMic
• qwaMixOARFocus
• qwaMixOARMaxReject
• qwaRECHalfDplxThresh

For the description of these parameters, please see the “QWA Parameters” section in the User's Guide.

API changes

The following lists all the API changes in this version:

• The default sample rate was changed to 8 kHz from 11 kHz in previous versions.
• Mute switches qwaSendMuteSwitch and qwaRecvMuteSwitch are not turned off when qwaReset() is called. In previous versions the mute switches were turned off when calling qwaReset().
• Enhanced the qwaInitialize() function to return QWA_ERR_INVALID_CHANNEL_CNT error for the following cases:
  • The microphone input count exceeds 4.
  • Receive input count equals 0 and any receive processing switch is on.
• The Voice Recognition mode was deprecated. For backward compatibility you can still set the parameter qwaOpMode to qwaVoiceRecog. Reading back the parameter qwaOpMode will always return qwaHandsFree now.
• The default value for the parameter qwaMaxFreqProc was changed from 4000 Hz to Nyquist Hz.
• The default value for the parameter qwaRefMaxDelay was changed from 0 to 200 ms if the qwaAutoDelaySwitch is idle or on.
• The FWHM minimum value for a PEQ node was changed to MAX(150Hz, f_center/10.0). This is equivalent to a maximum Q-value (Q = f_center/FWHM) of 10.0 for a center-frequency above 1.5 kHz.
• RCS API was enhanced to support post-initialization configuration to allow adjustment of parameters using binary configuration without restarting the target.
• New function qwaRcsCreateSerial() was added to the RCS API to support target connections through serial port.
• New function qwaSetConfig() was added to the QWA API to support post-initialization binary configuration.
• RCS error code QWA_RCS_ERR_SOCKET was renamed to QWA_RCS_ERR_COMM to indicate that an error has occurred during a serial port or TCP/IP socket operation.
• The default value for the parameter qwaRECSuppress was changed from 50 to 45.
• The default values for the qwaSendAGCAmpTgts and qwaRecvAGCAmpTgts parameters were reduced by 3 dB to [Min Max] = [-1500 -900].
• The default value for the qwaRecvBWEBrightness parameter was changed from 50 to 25.
• The default value for the qwaNRAtten parameter was changed from -1050 cdB to -1200 cdB.

Binaries

The following binary files are installed under $QNX_TARGET/, in subdirectories for each supported target platform:

ARMLE

• armle/lib/libqwa.a
• armle/lib/libqwaS.a
• armle/lib/libqwa_rcs.a
• armle/lib/libqwa_rcsS.a

ARMLE-v7

• armle-v7/lib/libqwa.a
• armle-v7/lib/libqwaS.a
• armle-v7/lib/libqwa_rcs.a
• armle-v7/lib/libqwa_rcsS.a

PPCBE

• ppcbe/lib/libqwa.a
• ppcbe/lib/libqwaS.a
• ppcbe/lib/libqwa_rcs.a
• ppcbe/lib/libqwa_rcsS.a

SHLE

• shle/lib/libqwa.a
• shle/lib/libqwaS.a
• shle/lib/libqwa_rcs.a
• shle/lib/libqwa_rcsS.a

x86

• x86/lib/libqwa.a
• x86/lib/libqwaS.a
• x86/lib/libqwa_rcs.a
• x86/lib/libqwa_rcsS.a

Header files

• usr/include/qwa_rcs_err.h
• usr/include/qwa_rcs.h
• usr/include/qwa_err.h
• usr/include/qwa_defs.h
• usr/include/qwa.h

The following binary files are installed under base_dir\install\qnxaviage-ap\2.0\ on Windows systems:

• QWALive\QWALive.exe

Sample source code

The following source code files with examples of how to use QNX Aviage Acoustic Processing 2.0 are installed under $QNX_TARGET/:

• etc/readme/qnxaviage-ap/examples/qwa_hf_example/qwa_hf_example.c
• etc/readme/qnxaviage-ap/examples/qwa_rcs_example/qwa_rcs_example.c

Fixed issues

Acoustic Processing 2.0 Rev. 1

• Fixed issue with link error for QNX AAP 2.0 for ARMLE-v7 variant. (Ref# 144147)

Acoustic Processing 2.0

• Fixed issue when diagnostic click did not scale with sample rate. (Ref# 82720)
• Fixed issue when output level for QWA_DIAG_WHITE_BAND_NOISE of qwaDiagPlaySignal depends on sample rate. (Ref# 65276)
• Fixed issue with QWALive controls not being visible when the display resolution is set to 1024x600 pixels. (Ref# 70975)
• Fixed issue where the soft limiter (SDL) malfunctions if clipping occurs near the beginning of a frame shift. (Ref# 74914)
• Fixed issue with latency slider in QWALive not showing the correct values after reinitialization. (Ref# 67330)
• Fixed issue where the receive side signal upsampling in mixed mode creates noise near Nyquist frequency. (Ref# 74952)
• Fixed issue where quantization of the Recv-In signal in conditions of low Recv-In signal levels can freeze send AGC for several minutes. (Ref# 72952)
• Fixed issue where the click test that is run at the beginning of the distortion test did not take the offset value into account. (Ref# 98611)
• Fixed issue where the wind-block algorithm was taking too long to adapt to an environment where wind buffet was not present causing speech quality degradation. (Ref# 94502)
• Fixed issue in QWALive tool where too many cue markers were inserted for dropped packets during audio streaming. (Ref# 67931)
• Fixed issue in QWALive tool where the streaming was continuing even if the library was reset causing audio artifacts in the wave file. (Ref# 68577)
• Fixed issue in QWALive tool where diagnostics cannot play 5 minutes signal. (Ref# 96038)

Known issues

• When the true system latency and the AEC latency as set by qwaRefDelay are significantly different, qwaAECEchoAlign can return values that are misleading. It can be in the wrong direction, or it can return a value close to zero when the true error is very large. (Ref# 103182)
• The wind-block algorithm is temporarily over-aggressive during sudden and large noise increase. (Ref# 75390)
• Temporary echo artifact during sudden radio volume increase. (Ref# 63210)
• Wind buffet suppression disabled when qwaWNDBSuppress set to 0. (Ref# J427799)

Benchmarks

Here's a summary of the abbreviations used in this section. For more details, see the Acoustic Processing User's Guide.

We used the following hardware for our benchmarks:

ROM Benchmarks

CPU Load/RAM Usage Benchmarks

NB single channel mode, AEC on, NR on

• DL-In/Mic-In = 8 kHz SR
• Recv-Out/UL-Out = 8 kHz SR
• MFP = 4000 Hz
• Base = AEC + NR
• AEC+ = Base + ADC + DPEQ + SDL + AGC + WNDB + RPEQ + RAGC + RDLC + RENS + RDTF
• +BWE = AEC+ + BWE
• +LFR = AEC+ + LFR

NB dual channel mode, AEC on, NR on, MIX on

• DL-In/(Mic-In)x2 = 8 kHz SR
• Recv-Out/UL-Out = 8 kHz SR
• MFP = 4000 Hz
• OAR = NR + MIX + OAR
• Base = AEC + NR + MIX
• AEC+ = Base + ADC + DPEQ + SDL + AGC + WNDB + RPEQ + RAGC + RDLC + RENS + RDTF
• +BWE = AEC+ + BWE
• +LFR = AEC+ + LFR

Mixed WB/NB single channel mode, AEC on, NR on, HFE on

• DL-In = 8 kHz SR
• Mic-In = 16 kHz SR
• Recv-Out = 16 kHz SR
• Mic-Out = 8 kHz SR
• MFP = 4000 Hz
• Base = AEC + NR + HFE
• AEC+ = Base + ADC + DPEQ + SDL + AGC + WNDB + RPEQ + RAGC + RDLC + RENS + RDTF
• +BWE = AEC+ + BWE
• +LFR = AEC+ + LFR

Mixed WB/NB dual channel mode, AEC on, NR on, MIX on, HFE on

• DL-In = 8 kHz SR
• (Mic-In)x2 = 16 kHz SR
• Recv-Out = 16 kHz SR
• Mic-Out = 8 kHz SR
• MFP = 4000 Hz
• OAR = NR + MIX + OAR
• Base = AEC + NR + MIX + HFE
• AEC+ = Base + ADC + DPEQ + SDL + AGC + WNDB + RPEQ + RAGC + RDLC + RENS + RDTF
• +BWE = AEC+ + BWE
• +LFR = AEC+ + LFR

WB single channel mode, AEC on, NR on

• DL-In/Mic-In = 16 kHz SR
• Recv-Out/UL-Out = 16 kHz SR
• MFP = 8000 Hz
• Base = AEC + NR
• AEC+ = Base + ADC + DPEQ + SDL + AGC + WNDB + RPEQ + RAGC + RDLC + RENS + RDTF
• +LFR = AEC+ + LFR

WB dual channel mode, AEC on, NR on, MIX on

• DL-In/(Mic-In)x2 = 16 kHz SR
• Recv-Out/UL-Out = 16 kHz SR
• MFP = 8000 Hz
• OAR = NR + MIX + OAR
• Base = AEC + NR + MIX
• AEC+ = Base + ADC + DPEQ + SDL + AGC + WNDB + RPEQ + RAGC + RDLC + RENS + RDTF
• +LFR = AEC+ + LFR

CPU Load benchmarks were calculated when running the library alone on the platforms listed above. CPU Load benchmarks may vary with the audio input to the library. Benchmarks for the 11025Hz, 22050Hz, 32000Hz, 44100Hz supported sample rates, are available upon request.

Processing Time

The QNX AAP 2.0 HF processing introduces the following delay in the Mic-In to Mic-Out (Mic-path) and Recv-In to Recv-Out (Recv-path) audio paths:

The t_proc value represents the time duration for processing one frame of hands-free audio with the qwaAudioIn() and qwaAudioOut() function calls (duration depends on CPU speed).

Documentation

The functionality provided by this package is described in the QNX Aviage Acoustic Processing User's Guide, which you'll find in the IDE's help system on Linux and Windows systems. On self-hosted Neutrino systems, you can find the documentation in the Photon helpviewer.

You'll find the QWALive User's Guide installed as base_dir\install\qnxaviage-ap\2.0\QWALive.pdf on Windows systems, where base_dir is where you installed QNX Momentics or QNX SDP.

You'll find the sample programs and a PDF version of the Acoustic Processing User's Guide in the $QNX_TARGET/etc/readme/qnxaviage-ap directory.

Technical support

To obtain technical support for any QNX product, visit the Support area on our website (www.qnx.com). You'll find a wide range of support options, including community forums.