Even if you know nothing about speech recognition
You may have realized something now.
The overwhelming success of speech-enabled products like Amazon Alexa has proven that some degree of speech support will be an essential aspect of household technology for the foreseeable future.
In other words, speech-enabled products would be a game changer as that offer a level of interactivity and accessibility that few technologies can match.
No GUI needed.
No texting needed.
Check out what books helped 20+ successful data scientists grow in their career.
No Emojis needed.
It’s all about speed.
Speed is a big reason voice is poised to become the next major user interface. Each decade, we’ve embraced a new way to interact with technology. We’ve evolved from character mode to a graphical user interface, to the web, to mobile.
Voice now offers a faster and easier way to communicate and accomplish tasks than mobile apps.
We can either tell Alexa what we need (turn off the lights, adjust the thermostat, and set an alarm — or all of the above using a single utterance like “Alexa, good night”), or you can pull your phone, unlock it, open the right app, and perform the task or tasks.
When you consider habitual use cases — those that keep customers coming back over time — the efficiency gained through voice adds up over time.
“Texting will decline in the future because of Alexa”
— Gary Vaynerchuk
Gary Vaynerchuk: Voice Lets Us Say More Faster
Therefore, that made me very interested in embarking on a new project to build a simple speech recognition with Python.
And of course, I won’t build the code from scratch as that would require massive training data and computing resources to make the speech recognition model accurate in a decent manner.
Instead, I used Google Speech Recognition API to perform the speech-to-text tasks with Python (check out the demo below which I showed you how the speech recognition worked — LIVE!).
By the end of this article, I hope you’ll have a better understanding of how speech recognition works in general and most importantly, how to implement that using Google Speech Recognition API with Python.
Trust me. It’s that simple.
Feel free to check out the source code here if you’re interested.
Let’s get started!
Why Google Speech Recognition API?
You may be wondering, “Is this the only API available given the growing demand and popularity of speech recognition?”
The answer is that there are also other APIs available either for free or paid services as below:
recognize_bing(): Microsoft Bing Speech
recognize_google(): Google Web Speech API
recognize_google_cloud(): Google Cloud Speech – requires installation of the google-cloud-speech package
recognize_houndify(): Houndify by SoundHound
recognize_ibm(): IBM Speech to Text
recognize_sphinx(): CMU Sphinx – requires installing PocketSphinx
That means you can get started right away without having to get authentication with either an API key or a username/password combination for other APIs.
However, the convenience of Google Web Speech API also comes with certain limitations: The API quota for your own keys is 50 requests per day, and there is currently no way to raise this limit.
This fits our use case if we just want to use this API for experimentation purposes. Note that if you’re running an app or a website that’s calling the API consistently, then you may need to consider getting a paid service from either of the APIs above.
Building speech recognition with Python using Google Speech Recognition API
To avoid boring you with technical details on how speech recognition works, you can read this great article that talks about the mechanism in general and how to implement the API.
In the following writing, I’ll show you how I implemented this API step-by-step by following the article.
But first, you need to install SpeechRecognitionlibrary using
pip install SpeechRecognition.
And we can use the Google Web Speech API that comes from this library itself.
In this implementation, I recorded my voice using own microphone and SpeechRecognizer accessed the microphone (Install PyAudio package to access the microphone) and recognized my voice accordingly.
Check out the code snippet below to understand the full implementation as they are relatively self-explanatory.
Function to recognize speech from microphone
To handle ambient noise, you’ll need to use the
adjust_for_ambient_noise()method of the
Recognizer class in order for the library to recognize your voice.
After running the
adjust_for_ambient_noise()method, wait for a second and let it analyze the audio source collected in order to handle ambient noise and capture correct speech.
Lastly, we need to implement
try and exceptblock to handle errors such as when the API is unreachable or unresponsive after sending requests, or when our speech is unrecognizable.
To use the function above, you can just implement the block below and…Voilà! You did it! ?
A simple demo on using Google Speech Recognition API
Now that we have the full implementation code ready. It’s time to see how this thing works.
I recorded a short video to show you exactly how the API works from recording my voice to returning that in text format.
Although this may not seem to be too accurate as what we may have expected, this is definitely worth the time to play around with the code and API!
Thank you for reading.
I hope you now have a better understanding of how speech recognition works in general and most importantly, how to implement that using Google Speech Recognition API with Python.
Feel free to check out the source code here if you’re interested.
I’d also recommend you to try out other APIs to compare the speech-to-text accuracy.
Despite the fact that speech-enabled products are not being widely used in businesses and our day-to-day life at this stage, I truly believe that this technology will disrupt a lot of businesses and how consumers will use products with voice recognition functionalities, sooner or later.
As always, if you have any questions or comments feel free to leave your feedback below or you can always reach me on LinkedIn. Till then, see you in the next post! ?
About the Author
Admond Lee is known as one of the highly sought-after data scientists and consultants in helping start-up founders and various companies tackle their problems using data with strong expertise in data science consulting and industry knowledge.
You can connect with him on LinkedIn, Medium, Twitter, and Facebook or book a call appointment with him here if you are looking for data science consulting for your company.Book online with Admond Lee
Admond Lee is known as one of the highly sought-after data scientists and consultants in helping start-up founders and…bit.ly
& AI for Education 2019 Initiative Committee Member
Data Science Consulting | Data Scientist | Speaker | Top Writer @Medium