Wallhanger
Posts: 8
Joined: Tue Aug 27, 2013 5:43 am

How well can Pi do speech recognition and synthesis

Tue Aug 27, 2013 5:50 am

Hello everybody. I was wondering how good the Pi is when it comes to speech recognition and synthesis. How many times do I have to say the words to make the Pi recognise what I'm saying. And does the voice sound robotic for synthesis?

tonywaite
Posts: 30
Joined: Sat Sep 15, 2012 10:13 am

Re: How well can Pi do speech recognition and synthesis

Wed Aug 28, 2013 10:47 am

Hi Wallhanger,

Espeak works easily and fabulously on the Pi for speech synthesis.

It is surprisingly good!

T o n y

User avatar
Jim Manley
Posts: 1600
Joined: Thu Feb 23, 2012 8:41 pm
Location: SillyCon Valley, California, and Powell, Wyoming, USA, plus The Universe
Contact: Website

Re: How well can Pi do speech recognition and synthesis

Thu Aug 29, 2013 2:02 am

If you want iOS Siri-like speech recognition of any arbitrary text, you might want to offload the processing to something like Google's version of Siri-like cloud processing (if it were like word processing, could you delete and insert clouds? ;) ). There's a pretty good tutorial here:

http://blog.oscarliang.net/raspberry-pi ... -like-siri

Another Google-based setup is at the following link, and it notes that a blue PlayStation 3 Wireless Singstar Microphone can be used with its included USB wireless dongle that works with the Pi:

http://raspberrypi-hacks.com/11/control ... ce-control


If you only need to recognize short phrases, you may want to try using Carnegie-Mellon University's Sphinx speech recognition library (no off-board processing, but limited vocabulary) and the University of Edinburgh's Festival for text-to-speech, then check out:

http://hackaday.com/2013/08/11/voice-co ... ightwaverf


There are more techniques that a Google search for "Raspberry Pi speech recognition" will find.

Enjoy!
The best things in life aren't things ... but, a Pi comes pretty darned close! :D
"Education is not the filling of a pail, but the lighting of a fire." -- W.B. Yeats
In theory, theory & practice are the same - in practice, they aren't!!!

dmb
Posts: 38
Joined: Wed May 30, 2012 9:33 am

Re: How well can Pi do speech recognition and synthesis

Sun Oct 06, 2013 5:16 pm

Not much to add, other than that I've been using Festival to do speech synthesis on my fridge (Yeah, really LOL!) and while the pi is OK, it's a little slow to build the speech stream. I'd recommend the cmu_us_slt_arctic voice, quality is pretty good and it's not as robotic as some of the voices. My family are getting used to being told what's going on by the fridge and don't have to work out what it just said ! You'll need a good speaker to drive it, I tried loads, and this one was the best and nearly the cheapest of the lot: http://www.currys.co.uk/gbuk/audio/hi-f ... 0-pdt.html.

I'm interested in the social effects of making machines sound like humans (i.e. whether people get freaked out or accept them) - seems most of the time they just think it's clever but sad ha ha !! Oh well ... just means it's not doing important enough things then :D

Heater
Posts: 12660
Joined: Tue Jul 17, 2012 3:02 pm

Re: How well can Pi do speech recognition and synthesis

Sun Oct 06, 2013 5:27 pm

I'm interested in the social effects of making machines sound like humans
I suggest you start by watching every episode of "Hitch Hikers Guide to the Galaxy" and "Red Dwarf".
Speaking lifts and toasters driving every body nuts.

Just now my girl friends laptop has some Yankee female voice announcing that some anti-virus package has been up dated on every reboot and there are some problems that need attention. Makes me want to throw the thing at the wall.

Festival with the Scottish accent is fun.

User avatar
pluggy
Posts: 3635
Joined: Thu May 31, 2012 3:52 pm
Location: Barnoldswick, Lancashire,UK
Contact: Website

Re: How well can Pi do speech recognition and synthesis

Mon Oct 07, 2013 2:12 am

Heater wrote:
I'm interested in the social effects of making machines sound like humans
I suggest you start by watching every episode of "Hitch Hikers Guide to the Galaxy" and "Red Dwarf".
Speaking lifts and toasters driving every body nuts.

Just now my girl friends laptop has some Yankee female voice announcing that some anti-virus package has been up dated on every reboot and there are some problems that need attention. Makes me want to throw the thing at the wall.

Festival with the Scottish accent is fun.
Lemme guess

Avast......
Don't judge Linux by the Pi.......
I must not tread on too many sacred cows......

User avatar
LemmeFatale
Posts: 253
Joined: Fri Feb 01, 2013 8:47 pm
Location: UK

Re: How well can Pi do speech recognition and synthesis

Mon Oct 07, 2013 12:12 pm

pluggy wrote:Lemme guess

Avast......
What a crime... They should be using a synthesised pirate voice, clearly.
Classic - Raspberry Pi Model B (512MB) with Motorola Atrix Lapdock
Lemcon-One - Raspberry Pi Model B (256MB) PiMAME TV-Box

PiGraham
Posts: 3553
Joined: Fri Jun 07, 2013 12:37 pm
Location: Waterlooville

Re: How well can Pi do speech recognition and synthesis

Mon Oct 07, 2013 1:03 pm

tonywaite wrote:Hi Wallhanger,

Espeak works easily and fabulously on the Pi for speech synthesis.

It is surprisingly good!

T o n y
Do you get no problems with longer text strings? I find that anything over ~2 seconds the speech breaks up and sounds like a Dalek.

This is OK:

Code: Select all

espeak "This is a short line of text."
This goes wrong after "text":

Code: Select all

espeak "This is a long line of text that causes a Dalek effect on espeak on the Pi.
What is your system configuration?

User avatar
scruss
Posts: 2265
Joined: Sat Jun 09, 2012 12:25 pm
Location: Toronto, ON
Contact: Website

Re: How well can Pi do speech recognition and synthesis

Mon Oct 07, 2013 1:14 pm

Heater wrote:Festival with the Scottish accent is fun.
That's no surprise, if you've ever heard Alan Black (one of the Festival lead developers) speak ...

I like Flite; it's small, fast, and only sounds a bit Scottish.
‘Remember the Golden Rule of Selling: “Do not resort to violence.”’ — McGlashan.

User avatar
DavidS
Posts: 4225
Joined: Thu Dec 15, 2011 6:39 am
Location: USA
Contact: Website

Re: How well can Pi do speech recognition and synthesis

Thu Oct 10, 2013 12:16 am

Depends on your OS and requirements.

I can not say much for Recognition as I do not know of a RISC OS one that will run in 32bit addressing mode.

As to synthesis there are a few options, and on RISC OS they are all greate, no studdering, or unwanted pauses. Just clear speach (In some cases the clearest synthesized speach I have heard on any system [excepting the Propeller]). I do not know about the options mentioned above (I have not heard of them).

On Linux I do not know of any as I do not play with speach on Linux at all. Same goes for BSD, Plan 9, Haiku OS, AROS, and any other OS.
RPi = The best ARM based RISC OS computer around
More than 95% of posts made from RISC OS on RPi 1B/1B+ computers. Most of the rest from RISC OS on RPi 2B/3B/3B+ computers

Return to “General discussion”