Chirp Tutorial: How to use Google's Chirp speech-to-text AI model on Google Cloud console.

Introduction

Chirp is Google Cloud's 2B-parameter speech model built via self-supervised training on millions of hours of audio and 28 billion sentences of text spanning 100+ languages. Chirp delivers 98% speech recognition accuracy in English and over 300% relative improvement in several languages with less than 10M speakers.

What we are going to do?

In this tutorial, we will learn how to effortlessly set up Google Cloud console to leverage the powerful Chirp speech-to-text AI model. This comprehensive tutorial provides step-by-step guide for an easy setup process and a quick start to use Chirp speech-to-text AI model. Sit back, relax, enjoy the tutorial and don't forget to make a cup of coffee ☕️.

Learning outcomes

• How to use Google Cloud console.
• How to set up Google's Chirp speech-to-text AI model on Google Cloud console.

We will cover the following steps:

1. Creating a Google Cloud account.
2. Creating a new project on the Google Cloud console.
3. Enabling the Speech API.
4. Creating an STT (Speech-to-Text) Recognizer using the Chirp model.
5. Creating a new Workspace for the project.
6. Performing a transcription on an audio file.
7. Viewing and downloading the transcription results.

Prerequisites

Nothing! Just a cup of coffee ☕️ and a laptop 💻.

Getting started

Step 1: Create a Google Cloud account

First, we need to create a Google Cloud account. If you already have one, you can skip this step and go here. If you don't have one, you can create a free account go here.

Step 2: Create a new project

Next, we need to create a new project. On the top left corner click on the project dropdown menu > New Project.

Give your project a name and click Create.

Step 3: Enable API

Go to Speech in Google Cloud console. Click ENABLE API.

Step 4: Create an STT Recognizer

Go over left sidebar navigation and click Recognizers > CREATE RECOGNIZERS.

Give your recognizer a name chirp-recognizer. Select Chirp as the model. Select the language en-US in our case, leave the rest of the settings as default and click Save.

Step 5: Create a new Workspace

Now, we need to create a new Workspace for our project. Go to Workspace dropdown menu and click New Workspace. You should see a sidebar pops up on the right side of your screen.

From the Create a new workspace navigation sidebar, click Browse > Create a new bucket. Give your bucket a name chirp-bucket and click Continue.

Note: Rest of the settings are optional, you can leave them as default.

Click Create and you should see that the new bucket is created.

Then, click Select > Continue > Create to finish creating your workspace for the speech-to-text UI.

Perfet! Now, we have a workspace for our speech-to-text UI. Let's move on to the next step.

Step 6: Create a new Transcription

Let's perform a transcription on our actual audio file.

Go over left sidebar navigation and click Transcription > New Transcription. From the New Transcription page, select your audio file through either uploading audio files Local upload or specifying an existing Cloud Storage file Cloud storage.

In this tutorial, we will use the Local upload option with a short English audio file of your own.

Note: The UI will assess your audio file parameters automatically, and you can also specify them manually.

Click Continue.

By default API version is set to V1, change it to V2, specify the language English (United States) - en-US in our case spoken in the audio file from the dropdown menu, then choose Chirp as transcription model and our newly created chirp-recognizer as the recognizer.

Click Submit and wait for a few seconds.

Step 7: View Transcription Results and Download

To see the transcription results, click the name of your transcription.

You may also download the transcription results in 4 different formats: JSON, TXT, SRT, and CSV. Let's download the transcription results in TXT format. Click Download > TXT > `Download.

Wrapping Up

This comprehensive guide has led you through implementing Google Chirp's speech-to-text AI model on the Google Cloud console. With step-by-step directions, you can now tap into Chirp's capabilities for precise speech recognition.

This AI tutorial aimed to provide a user-friendly, detailed roadmap for newcomers and experienced Google Cloud users alike.

Upon completing this tutorial, you should have the know-how required to apply Google Chirp's speech-to-text AI model effectively. Dive into Chirp's potential, incorporating it into projects and applications, and experiment with various languages and audio files.

AI hackathon ideas

Speech-to-text is a common starting point for AI hackathon projects such as voice notes, meeting summaries and captioning tools. Put your newfound knowledge to the test at one of the upcoming AI hackathons from Starfield Labs.

Frequently asked questions

What does the tutorial cover? It covers creating a Google Cloud account and project, enabling the Speech API, creating a Chirp recognizer, creating a workspace, running a transcription on an audio file, and downloading the results.

Which output formats can I download? The tutorial lists four: JSON, TXT, SRT, and CSV. It downloads TXT.

Which API version and model settings does it use? It changes the API version from V1 to V2, selects Chirp as the model, and uses the chirp-recognizer it created, with the language set to en-US.

How this tutorial was tested

This content was carried over from an earlier tutorial and updated for Starfield Labs. It was not re-run in this build, so no step has been tested here. Following it needs a Google Cloud account with a project, the Speech API enabled, and a Cloud Storage bucket, which may require billing to be set up. None of these was used.