Speech Services

Introduction

Speech Services is a cloud-based speech recognition service provided by 2RC Software.

Pay-As-You-Go: With a free account that requires no monthly subscription users can purchase time credits that allow audio to be submitted for speech recognition.

2RC enhances the already unparalleled accuracy of Google's speech engine with features that include:

• Phonetic Spelling: Use phonetic spelling (alpha, bravo charlie, ...) for accurate recognition of names, license plates, policy numbers and other non-standard words.
• Substitutions: Join pre-defined substitution lists or create custom substitutions to enhance recognition results and reduce editing time.
• Phrases: Provide a custom list of words and phrases that you use regularly for improved recognition results.

Requirements

• Browser running in Windows 7/8/10 or Windows Server 2008/2012/2016

Quick Start

1. Create a new account. Your username and password will be required to use 2RC Speech Services.
2. Purchase speech credits.
3. Use the on-line Speech Services to upload audio files to be converted to text.

Submitting Audio

You can submit audio to be converted to text in one of three ways.

1. On-Line Speech Services: This online service allows you to upload an audio file and have the resultant text returned. You are required to enter your username and password in order to access the credits in your account. When running in a browser in Windows, once the text is returned you can click the copy-to-clipboard icon, and then paste the text in whatever application you would like - Word, an EMR form, etc.
   
   

   

   
   
2. Media Player: Media Player is a free, Windows desktop application that can play a number of audio formats and provides integrated support for performing speech recognition using your 2RC Speech Services account. Download Now


3. Transcribe Assistant: Transcribe Assistant is a Windows desktop productivity tool for professional transcriptionists. It has built-in support for submitting audio files to 2RC Speech Services.

Your Account

An account is required to use 2RC Software's speech recognition services. Your account keeps track of how many speech recognition credits you have available. One credit equals one second of audio. Therefore, in order to submit a one minute long audio file for recognition, you will need 60 credits in your account. You can create a new account for free.

When you have an account, you can purchase speech credits. When you submit audio files for recognition your credits will be reduced one credit per one second of audio processed.

Organizations

An Organization is a high level account that contains personal accounts. Using an organization to contain other accounts allows a group of people to share a common pool of speech credits as well as share common substitution and phrase lists. Each Organization is assigned one of the personal accounts to act as the administrator. The administrator manages the contacts/accounts, groups, phrase lists, and organization lists.

Manage Your Organization

Phonetic Spelling

The phonetic spelling feature provides a means of accurately spelling out a name, vehicle plate number, insurance policy number, and other strings of alphabetic and numeric characters that aren't regular words. Phonetic spelling is initiated by saying "Start Spell", and then speaking the phonetic words that represent the characters and numbers you wish to have printed. End the sequence by saying "Stop Spell".

For example, speaking...

Results in the text:

The Phonetic Spelling feature is available by default to all users.

Phrases

A list of words and phrases that provide hints to the speech recognition engine.

A word or phrase can assist in speech recognition results if there is a good chance it will be used in a recorded dictation and/or it is a word not likely to already exist in the default dictionary of words. Despite the very large default vocabulary, your dictated audio may contain proper names or domain-specific words. These may be good candidate words for adding as a phrase.

Multi-word phrases boost the probability of recognizing those words in sequence but also, to a lesser extent, boost the probability of recognizing portions of the phrase, including individual words.

1. Each phrase must be less than or equal to 100 characters.
2. The total number of phrases that can be provided to the speech recognition engine is 500.
3. The total combined number of characters must be 10,000 or less.

Phrases belong to Groups and are managed by logging into Organization Management. Phrases and Substitutions belonging to 2RC Groups cannot be modified.

Substitutions

Substitutions are used to correct common recognition mistakes, fix common case issues, and to expand or contract abbreviations.

Before:

After:

Substitutions belong to Groups and are managed by logging into Organization Management. Phrases and Substitutions belonging to 2RC Groups cannot be modified.

Strategies for Handling Case

Say you want to substitute the word "color" with the British/Canadian spelling "colour". In some cases the original word may begin with a capital letter and in other cases the word appears mid-sentence and therefore begins with a lower-case letter. There are two techniques:

1. Use two substitutions and set "case Matters" flag to true.
   
   color,colour,true
   Color,Colour,true
   
   
2. Leave of the first letter of the word and only search for and substitute for the end of the word.
   
   olor,olour,false
   

Command Reference