Hi
I need to do some voice recognition with ARM based microcontrollers. I like to have a system with about 20 to 50 words dictionary and when user says those words the system can recognize and do something in order to them.
So, is there any library of facility for that?
How can I do it and I should search what keywords to find any solution?
Thanks in advanced!
HI Masoomi and Jens,
The company Sensory has exactly what you are looking for. They have a small vocabulary embedded voice recognition engine called "Truly Handsfree" which does exactly what you want. It can run on the ARM Cortex-M4.
-Paul
That's awesome.. Now things begin to get interesting....
Combining Voice Recognition with WiFi, TFT displays, SD cards, camera modules and other hardware (status sensing), can result in very cool gadgets for home automation.
Hisorry if i'm late and thanks about suggestion
Hello and thank you
But I think it's commercial and a bit large foe my project. I want something open-source and free, also light. and I like to implement it to my mother tongue Persian.
Hi masoomi and welcome to the community.
This sounds really interesting .
Unfortunately, I do not know much about voice recognition, but I would expect that a Cortex-M4 or Cortex-M7 would be able to do the job.
The Cortex-M4 and Cortex-M7 both have a DSP-extension in the instruction set; this would make jobs like voice recognition spend less CPU time on data processing.
Maybe pbeckmann would be able to add a few notes (he's the founder of a company called DSP Concepts, which have a very interesting product called AudioWeaver).
Maybe Paul can tell us if there there is a voice recognition module for AudioWeaver.
Hi,
Like the softwares on Linux are open source, you can always try to recompile directly on Arm plateform . Recompilation is not always necessary when using an interpreted scripting language such as Python.
I suggest you watch it (to rebuild)
http://sourceforge.jp/projects/julius/downloads/60273/julius-4.3.1.tar.gz/
Or it, in pythonSpeechRecognition 3.4.6 : Python Package Index
NXP make an evaluation kit called OM13090 which is an LPC54114 audio and voice recognition Kit.
Search result for : OM13090 - Future Electronics
I recently saw a speech recognition project by Mozilla and it seems they use this: PocketSphinx
Arm just released an open source library for key word spotting:
https://github.com/ARM-software/ML-KWS-for-MCU
I suggest you try free audio to text converter. It supports different languages and provides automatic and manual transcription services. Moreover, it gives you 30 minutes for free. I hope, it will be helpful.
Latest repo for KWS on Cortex-M based platforms:
github.com/.../tflu-kws-cortex-m
You can use raspberry pi for that. Also, for compiling, use the GCC (its a C compiler). There's a tutorial as well regarding this compiler, google it.
You can use Raspberry pi or STM32F411.