This page describes some of the things which can be done with audio on the SmartBadge version 4.
To set the samping rate:
set-sample-rate 8000
It is important to note that the sampling rate can only be set to certain values, thus the exact rates are not necessarily exactly what you ask for:
set-sample-rate -v Failed to set sample rate to 44100 actual speed is 43200 set-sample-rate -v -r 16000 selected audio rate is 16000 Failed to set sample rate to 16000 actual speed is 16045 set-sample-rate -v -r 8000 selected audio rate is 8000 Failed to set sample rate to 8000 actual speed is 8022
See set-sample-rate.c source code for details of the use of the ioctl(dspFD, SNDCTL_DSP_SPEED, &ioctlParam ), which does the actual setting of the sampling rate for the OSS audio driver.
cp /opt/Badge4/support/badge_fs/hello_world/cwbeep /tmp/cwbeep dd if=/tmp/cwbeep bs=1024 count=4 of=/dev/sound/dsp
dd if=/dev/sound/dsp bs=1024 count=20 of=/tmp/z2
dd if=/dev/sound/dsp bs=1024 count=20 of=/tmp/z2
For a more complex example, consider the modification to Jef
Poskanzer's "saytime - audio time hack for the SPARCstation", which
resulted in
saytime2.c
The most important changes were to change from of 8 bit uLaw (used by
the SUN) to 16 bit linear samples (used by the badge) and setting the
sampled rate for the output to 8000. The complete set of changes is
documented in the source file.
"aumix" by default uses /dev/mixer to make it work you either have to
To find out the current settings:
# aumix -d /dev/sound/mixer -q vol 65, 65 bass 50, 50 treble 50, 50 line 88, 88, R mic 88, 88, R
You can adjust the volume by:
# aumix -d /dev/sound/mixer -v 80:80 # aumix -d /dev/sound/mixer -v 90:90 # aumix -d /dev/sound/mixer -v 70:70
To find out all the options, run "aumix" without any arguments.
Then you can play an mp3 file by specifying its name:
mpg123 /opt/Badge4/maguire/xxx.mp3
Where /opt/Badge4/maguire/xxx.mp3 is an MP3 file.
Alternatively you can copy the MP3 file to /tmp and then play it from there to avoid an problems with your network connectivity while playing it.
Note that this section is based on e-mail I sent to [hp_badge4] with the subject: Text to Speech and temperature by voice on Mon, 29 Jul 2002 22:20:44 +0200 (MET DST).
If you would like to have text read to you, you can use the CMU Text To Speech synthesis programs:
flite -- text message or a file or a URL and flite_time -- generate a phrase version of the time value you give it.
There are systems such as CMU's Flite and University of Edinburgh's Festival:
"Flite (festival-lite) is a small, fast run-time synthesis engine developed at CMU and primarily designed for small embedded machines and/or large servers. Flite is designed as an alternative synthesis engine to Festival for voices built using the FestVox suite of voice building tools.
For further information see the web links from http://www-2.cs.cmu.edu/~awb/ and specifically: http://www.speech.cs.cmu.edu/flite/
The 16kHz sample rate version of Flite 1.1 is 5,900,220 bytes in size.
I downloaded the CMU Flite speech synthesis package (which is available from http://www.speech.cs.cmu.edu/flite/index.html). They already have a precompile ARM Linux version (which they ported to the iPAQ). To run it on the badge you fetch the file and unpack it:
zcat flite-1.1_bin16KHz_arm-linux.tar.gz | tar -xvf - # unpacks an executable "flite16k" # make the link /dev/dsp link so that it is happy ln -s /dev/sound/dsp /dev/dsp # with the 2.95.3 files unpacked as per "Using the GNU libraries" section # of the Badge4 Embedded Development Kit: Bastille Day Release document ln -s /opt/Badge4/2.95.3/arm-linux/lib/ld-linux.so.2 /lib/ld-linux.so.2 # define a path to the usual gnu libraries LD_LIBRARY_PATH=:/lib:/opt/Badge4/2.95.3/arm-linux/lib export LD_LIBRARY_PATH # you can now run it: for example: ./flite16k "Hi Chip! How are you? This is a test. 1, 2, 3, 4, 5 ..." # for something a bit fancier try: ./flite16k "The time is now `date +'%A %d %B %Y at %H:%M'`"
There is also a program specifically for time annoucements. Unpack with: zcat flite-1.1_time_bin16KHz_arm-linux.tar.gz | tar -xvf -
./flite_time `date +'%H:%M'` The time is now, exactly twenty past seven, in the evening.
As their documentation says, it is a scottish accent!
If you give the flite16k program a file name as an argument it will read you this file.
So I put the following text into a file "flite-description.txt".
Flite (festival-lite) is a small, fast run-time synthesis engine developed at CMU and primarily designed for small embedded machines and/or large servers. Flite is designed as an alternative synthesis engine to Festival for voices built using the FestVox suite of voice building tools. Flite 1.1-release is now released as source with a pre-built binary for ipaq Linux. Flite offers: Completely in C (no C++ or Scheme) for portability, size and speed Reimplentation of the core parts of the Festival architecture (HRG) allowing close compabilility between voices built for each system. Voices compiled into C (mostly automatically) from FestVox format voices Thread safe Scalable voice size with all data const so it can be in ROM Target architectures, ipaq (Linux/WinCE) and smaller Flite is in basically written and is in its first stages of testing before release, as free software. A small diphone voice based on the CMU KAL voice is included. along with a sample limited domain talking clock. Here are slides about Flite from a recent talk given by AWB slides.ps or slides.pdf . Here is a recent publication at the 4th ISCA Speech Synthesis Workshop by Alan W Black and Kevin A. Lenzo in html or postscript.
Then I had the system describe itself with:
./flite16k flite-description.txt
While it is not a great rendering, it is reasonably intelligible.
If you use the my new program report-temp-improved (file: http://www.it.kth.se/~maguire/report-temp-improved.c compiled with: arm-uclibc-gcc -o report-temp-improved report-temp-improved.c )
First install the badge sensors:
# modprobe badge4_sensorse
Then try it with:
# ./flite16k "`report-temp-improved`" # ./flite16k "`report-temp-improved -c`"
For more fun try:
# ./flite16k "The temperature is `report-temp-improved -c`" # ./flite16k "The temperature is `report-temp-improved`" # ./flite16k "The temperature is `report-temp-improved -cF`" # ./flite16k "The temperature is `report-temp-improved -c`" # ./flite16k "The temperature is `report-temp-improved -cFt`" # ./flite16k "The temperature is `report-temp-improved -ct`" # ./flite16k "The temperature is `report-temp-improved -Ft`"
Yes, I know it is a very expensive talking temperature device; but it is pretty slick how easy it is to make such an application. The next obvious application is probably a program that runs in the background and invokes a script when the temperature exceeds a certain setpoint.
To generate new voices for Flite, you can use the Festvox software, see http://festvox.org/