giselle22
Posts: 1
Joined: Tue Jun 02, 2020 5:08 am

speech to text raspberry pi offline ver

Tue Jun 02, 2020 5:13 am

can someone teach me step by step speech to text using raspberry pi offline ver, i want to create a project for people that having trouble with listening

gbafamily
Posts: 13
Joined: Sun Jun 07, 2020 9:55 am

Re: SPEECH TO TEXT RASPBERRY PI OFFLINE VER

Sun Jun 07, 2020 9:07 pm

Mozilla DeepSpeech is a Speech-To-Text engine that works offline (no Internet required) on a Raspberry Pi 4. See the following link to install it.

https://github.com/touchgadget/DeepSpeech

User avatar
recantha2
Posts: 364
Joined: Wed Nov 14, 2012 9:34 am
Location: Potton, Bedfordshire
Contact: Website Facebook Twitter

Re: SPEECH TO TEXT RASPBERRY PI OFFLINE VER

Mon Jun 08, 2020 10:49 am

Alternatively, take a look at eSpeak: https://www.instructables.com/id/Make-y ... -Pi-speak/
--
Michael Horne - @recantha on Twitter
Raspberry Pi Pod blog - http://www.recantha.co.uk/blog
Cambridge Raspberry Jam - https://camjam.me
Pi Wars - https://piwars.org

User avatar
rpdom
Posts: 18530
Joined: Sun May 06, 2012 5:17 am
Location: Chelmsford, Essex, UK

Re: SPEECH TO TEXT RASPBERRY PI OFFLINE VER

Mon Jun 08, 2020 11:42 am

recantha2 wrote:
Mon Jun 08, 2020 10:49 am
Alternatively, take a look at eSpeak: https://www.instructables.com/id/Make-y ... -Pi-speak/
That's for Text to Speech, not Speech to Text.
Unreadable squiggle

User avatar
recantha2
Posts: 364
Joined: Wed Nov 14, 2012 9:34 am
Location: Potton, Bedfordshire
Contact: Website Facebook Twitter

Re: SPEECH TO TEXT RASPBERRY PI OFFLINE VER

Fri Jun 12, 2020 8:42 am

rpdom wrote:
Mon Jun 08, 2020 11:42 am
recantha2 wrote:
Mon Jun 08, 2020 10:49 am
Alternatively, take a look at eSpeak: https://www.instructables.com/id/Make-y ... -Pi-speak/
That's for Text to Speech, not Speech to Text.
Oops! Sorry - was doing text to speech for a project and mis-read!
--
Michael Horne - @recantha on Twitter
Raspberry Pi Pod blog - http://www.recantha.co.uk/blog
Cambridge Raspberry Jam - https://camjam.me
Pi Wars - https://piwars.org

manfred-muc
Posts: 1
Joined: Sat Dec 26, 2020 10:07 pm

Re: SPEECH TO TEXT RASPBERRY PI OFFLINE VER

Sat Dec 26, 2020 10:14 pm

Hi, I am also very interested in a speech to text system for raspberry Pi ... in German.... If somebody is working on that, please let me know.
I am thinking about a handheld device with 7" display presenting what is spoken as text on the screeen...

Salohcin
Posts: 2
Joined: Sun Jan 31, 2021 4:05 pm

Re: SPEECH TO TEXT RASPBERRY PI OFFLINE VER

Sun Jan 31, 2021 4:57 pm

gbafamily wrote:
Sun Jun 07, 2020 9:07 pm
Mozilla DeepSpeech is a Speech-To-Text engine that works offline (no Internet required) on a Raspberry Pi 4. See the following link to install it.

https://github.com/touchgadget/DeepSpeech
This looks promising, thanks for the link. Do you have DeepSpeech running on RPi 4? I suppose I'll need to use the Lite OS to have enough room for it.
My first RPi project is a talking To-Do list using espeak for text-to-speech, and being able to speak my to-do list tasks into this device would be great!

openbsd98324
Posts: 17
Joined: Tue Jan 05, 2021 5:58 pm

Re: SPEECH TO TEXT RASPBERRY PI OFFLINE VER

Sat Feb 06, 2021 12:34 pm

It is possible to run dragon naturally speaking for qemu on the rpi 4 with windows.

qemu emulates Windows older version, but this is not opensource at all.

daphnee
Posts: 2
Joined: Thu Mar 04, 2021 10:02 am

Re: SPEECH TO TEXT RASPBERRY PI OFFLINE VER

Fri Mar 05, 2021 7:33 pm

You can install raspberry PI voice recognition with SOPARE. Here's how you can.

SOPARE stands for SOund PAttern REcognition and it is python based developed for raspberry Pi. You can process real time audio offline here. SOPARE is designed to identify the sounds learned under different situations. Mean you can trained word in different languages.

Hardware requirements for installation.

Obviously computer and SOPARE is developed for raspberry then we will go with this. You can use other software too if desired as it will runs on other software too. But be sure to use hardware with multi core processor like Pi 2 or 3. Note: SOPARE will not work with older raspberry hardware like Pi B or B+ as they do not has multi core processor. Also you will need power supply and micro SD card to run Pi.

Then you will need a microphone or USB mic. Microphone is an essential component here. Hence make use of right microphone suits you. For large distance and short distance recognition Cheap USB mic will work fine for you. Test your mic before installation and configuration here at mictest.me. It will help you to select best microphone that fits you.

That's enough with hardware now let talk about software requirements. First, SOPARE should run on every available raspbian version. Choose what you prefer.You need to download, install, and configure raspbian.

Now you have everything like hardware, mic and Operating system. In case of raspbian you has lists of software for further installation. Just install essential libraries manually with the following command.
sudo apt-get update
sudo apt-get install build-essential python-pyaudio python-numpy python-scipy python-matplotlib
Optional task: Create a development directory in your home directory. If you want to do this optional task execute following command.
cd
mkdir dev
cd dev
Now you are good to go to install SOPARE from github.
To be ready completely and to follow complete instruction we need two more directories.
cd sopare
mkdir tokens
mkdir samples
Here you will install SOPARE successfully. Carry a test to figure out if microphone is configures and is usable. To test audio start SOPARE and follow command given below.
python sopare.py -u
python test/test_audio.py
In case of no errors (if everything works fine) You will see this interface
sopare 1.5.1
starting unit tests...
...
unit_tests run successful!
done.

test_audio init...
... ALSA related information ...
testing different SAMPLE_RATEs ... this may take a while!

Your sopare/config.py recommendations:

SAMPLE_RATE = 48000
CHUNK = 512
THRESHOLD = 100
Now your are able to edit and change configuration and file as per recommendation.
nano config/default.ini
After you save the config its time to train the words. Example if you want to train word test then execute command given below
./sopare.py -v -t test
Start saying the word „test“ shortly after the line after below given line appears on the screen.

INFO:sopare.recorder:start endless recording

You should see lots of lines rush over your monitor. This is good as SOPARE logs some debug information. If SOPARE started the training before you said something, lines will start rushing because something triggered the THRESHOLD. In that case delete the trained file(s) and start the training again, maybe with a higher THRESHOLD.
Here is the command to delete the files and the dictionary and start again:
rm dict/*.raw
./sopare.py -d "*"
You can repeat the training round a few times. Normally 3 times is enough to get first results After the training SOPARE must create an internal dictionary from the training:
./sopare.py -c
You reached the end of installation process. To check if the trained words are recognized correct Start SOPARE in endless loop mode and say „test“:
./sopare.py -l
Depending on your mic, your environment, the count of test rounds and lots of other things you should see that SOPARE is able to recognize the word test as it appears on the screen in square brackets.
['test']
So that is the step by step guide for you. Hope this will works for you and you install successfully raspberry Speech to text hardware and software. Best of luck for your project too.

Unounited
Posts: 55
Joined: Wed Oct 21, 2020 1:48 pm

Re: speech to text raspberry pi offline ver

Thu Mar 25, 2021 1:57 pm

I used to have speech to text a couple years ago when I used my old raspberry pi, unfortunately I forgot what the commands were and how to use it, however I did remember how to install it. When this topic came, that just made me remember how to use speech to text and what the commands were.

Whoever made this topic is the one I am going to thank, because I loved using it in the past, now I can use it again.:D Cheers!

Return to “Assistive technology and accessibility”