symblai / Get-Spelling-Sequence

Get-Spelling-Sequence

The purpose of this package is to find a spelling sequence coming from a conversation. Letters phonemes are limited and are not easily understood by spelling a sequence of a word to humans or machines that are listening on the other line. In order to capture the letters spelled correctly saying the letter followed by a word increase acccuracy by captuaring the first letter in the word which has more phonemes with higher accuracy.

Pre-Requirements

• Symbl appSecret and appId - If you don't have a Symbl account for signup look it up in this Link.
• Orignal recorded file was processed in Symbl using Async Audio/Video API that was completed and generated a conversationId
• The participant who spelled the word sequence will do it either by using the structure <letter> as in <word> or <letter> like <word>. For example "A like apple", "B as in Boy".

How does this work?

1. The script will use the messages API with "verbose=true" to get a JSON response with the messages word level.
2. Each word in the sequense will look for "as in" or "like" and print the letter of the word that follows it.

How to use?

1. Download by using the command git clone https://github.com/symblai/Get-Spelling-Sequence.
2. Get into the downloaded folder using the command cd Get-Spelling-Sequence.
3. Add to .env your Symbl APP_ID and APP_SECRET and save the file.
4. Open index.py, modify conversationId and save the new changes.
5. Run the file python index.py and if found the spelled letter sequence will be printed.