interesting sound card for Picaxe robot projects

womai

Senior Member
Maybe this post would fit better under Audio/Visual or Robots, but then again, the Active Picaxe Forum gets the most views:

I remember past discussions about how to add sound/speech to a project, so maybe some may find the following interesting - I stumbled across it by chance:

http://www.talkbotbrain.com/
 

westaust55

Moderator
Seems that they are another Roman Black creation:
http://www.blackrobotics.com/Brains_for_robots_robotic_controllers.htm
Certainly using Roman Black audio file creation software.

If someone needs both sound and control up to 8 servo motors the Talkbot Brain module is certainly a great add-on module.

TalkBot Brain holds a total of 23.8 seconds of 1.5bit sound, indexed for playback of up to 256 individual sounds, all at a playback rate of 44.1kHz. That is about 50 spoken words!
Cost is US$46-27ea with US$14-80 Shipping


IMHO, the SOMO 14D is still a very worthwhile module where only audio playback is desired.
The SOMO 14D can use up to 512 audio files with a max of 2GB total capacity (max 1GB for a single file I ahve previously been informed) so total duration could reach around (512 * ~4 = ) 2038 minutes
Cost is US$25-00ea with US$?? Shipping (around AUD$10 here is Aust)
 

westaust55

Moderator
Stumbled across this and thought of this thread.
Speakjet @ Sparkfun: http://www.sparkfun.com/commerce/product_info.php?products_id=9578
The Sparkfun chip was also mentioned yesterday by boriz (the second link in this post):
http://www.picaxeforum.co.uk/showthread.php?\ p=122224


Sounds would not be a problem and since it has a 64-byte input buffer, the speakjet chip should be able to generate speech under PICAXE control.
From work I did with Phomene/Allophones years ago the duration of the individual sounds is of the oder of 0.5 to a few msec each. COuld take from1 or 2 up to 10 allophone (sounds) for a single words. If the buffer is pre loaded before starting to say words then a short sentence should be easily generated.

Text to speech generation requires a lot of program space (in PICAXE terms - 2kbytes on say a 6800/6502 microcontroller) and quality is often a function of the size of the exceptions dictionary.
I would foresee that under PIACXE control the user would have predefined sequences of allophones for desired words rather than disecting text and creating a string of allophones for that words.
 
Last edited:
Top