Introduction

The Ins8.ai Automated Hyperlocal Speech Recognition (AHSR) software comes with pre-built APIs for easy access and integration into speech application systems, without needing to replace existing audio recording infrastructure.

For more details on REST and WebSocket APIs, refer to the Ins8.ai API Documentation.

Specifications

To ensure compatibility of the Ins8.ai software into your speech system for on-premise deployments, ensure that the following criteria of your audio inputs and environment are met:

1) Audio format: WAV (PCM-16 bit, uncompressed).

2) Audio size limit: The audio size limit is determined by the configuration of your infrastructure.

3) Audio quality: The standard model of Ins8.ai AHSR is configured to process and transcribe speech signals similar to audio settings of a call center. Audio with noisier environments can lead to inaudible speech and errors in the transcription.

4) Language support: Languages spoken in the audio must be the same and/or within the capability of the chosen AHSR language pack.

5) Compatibility with different platforms: As the Ins8.ai AHSR licensed software is delivered simply as docker images, which are widely supported on different deployment platforms.

6) Security: Our Docker images have passed stringent security scans and are compliant with the latest cybersecurity regulation standards i.e Common Vulnerabilities and Exposures (CVE).

Steps

The method outlined below is one of ways to integrate Ins8.ai into your speech system of choice:

1) Determine the application (audio recorder, batch audio database, etc) in your system that would communicate and make the API request to our software.

2) Integrate the Ins8.ai software, delivered as docker images, into your pipeline.

3) From your pipeline, request for the Ins8.ai software to start on a batch or streaming transcription.

4) Once executed, our software takes in the audio bytes through REST API or WebSocket and converts them into text.

5) Ins8.ai returns the text transcription result as a JSON output.

6) You can now use the JSON output for any downstream automation, processes or services of your choosing.

Note that no audio or transcription data are stored in our software throughout the process.

Try Now

Ins8.ai provides a quick and easy test run of the software. Try it over at the Ins8.ai webapp, run our API on your IDE or quickly spin up an ASR webapp.

Authored by Danny Yap, Clement Tay and Renee Tan

06 Apr 2023