Watson Speech
Please also use the Watson Apps Community, and the Watson APIs Topic Group for collecting links to high quality technical enablement materials.
This is a “landing page” for Watson Speech services. This includes both Watson Speech-to-Text (STT) and Watson Text-to-Speech (TTS). Look here for links to high value technical content specific for this service.
References for Further Reading
- Why The Overall Voicebot Solution And User Experience Are More Important Than Speech Accuracy - Do you want to have a successful working chatbot, that talks and understands what your customers are saying? Read this article - great advice on setting appropriate expectations for quality and accuracy.
- Watson Speech to Text: How to Plan Your Migration to the Next-Generation Models - some guidance on how to migrate to the newer, more accurate, speech models now available for Watson Speech-To-Text (August 2021).
- Replacing My Old Interactive Voice Response (IVR) System with IBM Watson AI Technologies - A great not-too-technical blog post by Marco Noel on how to get from your current IVR nightmare, to a chatbot that your customers will find helpful and delightful.
- Speech to Text (STT) API Reference - The API reference page for STT.
- Text to Speech (TTS) API Reference - The API reference page for (TTS).
- TTS/STT Redbook - A large document with some best practices and an example use case using NLU. Very big, lots of details.
- IBM Watson Text to Speech: Neural Voices Generally Available - Read about and listen to the newer more natural sounding voices that are available as of July 2019.
- IBM STT Model Hits New Milestone - no technical information here, but an overview of how STT is done, and how IBM continues to improve it’s speech recognition models.
- Data Collection and Training for Speech Projects - great article with a step-by-step approach to training your speech based applications.
Code
- How To Control Background Noise with Watson Speech-To-Text - Find out how to optimize how well you are able to deal with noisy environments with your Watson Speech-To-Text service.
- Simple Node.js STT App - A simple starter app in Node.js that utilizes the Speech to Text service.
- Train a Speech-to-Text Model Code Pattern - a demo and associated GitHub repo with code which highlights how to build out a specialized model for a Speech-to-Text service.
- A Postman Collection for Training IBM Watson Speech to Text - a collection of Postman instructions and directions for using them, written by Peter Tuton, that you can use to customize a language model and an acoustic model for your improved Speech-to-Text service.
Testing
- New Speech Testing Utilities for Conversational AI Projects - cool blog post by Andrew Freed on how you can measure the accuracy of your Watson Speech transcription, and the overall “sound” of your chatbot. What are they going to sound like? Is that acceptable to you?
- New Python Scripts to Measure Word Error Rate on Watson Speech to Text - nice blog post by Marco Noel on how you can measure the accuracy of your Watson Speech model, with links to a GitHub project with a Pythin script that shows you how to do it.
- Testing Strategies For Speech Applications - make sure that you are testing your speech applications adequately. Andrew Freed has some good approaches to doing testing in this area - it is not as easy as it sounds (pun intended).
- Announcing the Release of the Voice Agent Tester - a testing resource to use for testing and improving your voice agent over time. It uses Watson Voice Gateway and Watson Speech services to test out your solution.
Integrations
- Wave Hello to Watson Assistant Voice Interaction, and Goodbye to Complex Phone Trees - nice article by Preeth on using the Watson Assistant for Voice Interaction (WAVI) integration to eliminate phone trees and annoying IVR systems.
- Quickly improve your voice agent with a speech model - a great way to rapidly kickstart the use of a custom speech model to help make your chatbot really chat….
Customizations
- Watson Speech Services: Introducing a Simple Interface to Train STT and TTS Custom Models! - great article with links to a GitHub repo that has a simple UI that can help you customize your Watson Speech services more easily.
- Watson TTS Python Tool With Examples - another GitHub repository, this one has some examples of code that will do different things with Text-To-Speech, so you can see how to do some commonly requested things with the Watson Speech services and Python.
- Tune by Example: How to Tune Watson Text to Speech for Better Intonations - nice article on how you cane tune the way that your Watson Text-To_Speech model talks. Don’t settle for “robotic” and choppy voices, have your responses sound natural and clear.
- Why The Audio Compression Format Impacts the Speech to Text Transcription Accuracy - Great exploration of how the recording format of speech files has a direct impact on the quality of the training of your speech models.