Conduit Transcriptions

With over a decade of providing transcription services to private and public parties, Conduit Transcriptions was relying on an outdated legacy platform that needed upgrades. The client continues to serve a wide range of clients like Federal agencies and private law firms. These firms count on Conduit to provide secure transcripts quickly and accurately. Throughout the years, their systems have seen a downturn in speed and reliability. Their outdated platform was in need of enhancements, but upgrading was no easy task. Solving Conduit’s problems relied on customized developments to cloud-based speech recognition software (Rev.ai) that can introduce and expedite the work process inside a custom user interface.

Conduit has constantly met client expectations with quality work throughout their years in business, however, it was clear they needed a faster workflow to meet their goals of future growth. Their main service of providing official transcripts includes turning speech to text files that are downloaded into an API and further examined by administrators. Administrators then edit text and titles into fully legal and formal documents to be returned to clients for official record keeping. The interesting, and challenging, focus here is that the files originated from audio records. And, converting thousands of conversations into accurate, formal documents needed custom technology to surmount the different task levels more efficient.

The process of converting audio files and integrating speech recognition software into a custom web application presented many hurdles for Conduit. The previous process contained many steps:

1. An assignment starts with speech recognition cloud technology converting MP3 files into a Transcription Page.

2. Once processed, the Transcription Page is made available to managers who delegate the assignment to proper personnel.

3. Conduit personnel and translators edit the text that’s misinterpreted by the software – Important to note, Transcription Pages can contain hundreds of pages. 

4. Before finalizing, all titles, case numbers and client agency names needed to be filled in properly, line by line.

5. An output file is generated from these edits and a “Super Admin” is notified. 

Through each of these tasks, notes, updates and check-ins are processed between managers and administrators.This time consuming process depended on making edits as quickly, and as accurately as possible, while communicating with little space for errors. Conduit understands that any speech-to-text technology is not 100% accurate, but different levels of inaccuracy exist, and those needed to be minimized while also automating simple tasks to be completed.

For Conduit to be able to grow and take on more clients, nearly all stages of their workflow needed to be updated. This included updating their API integration software connected to third party speech recognition software, the output quality of Transcription Pages, the dashboard for users and managers, and security controls to protect records and client data. 

Previously, many errors were present when MP3 files were converted to text. Now, with the upgrade, the speech recognition software outputs data in JSON (Javascript Object Notation) files that are easy for people to read and write but also easy for machines to generate and analyze, making more accurate speech-to-text files. This change reduced the amount of editing needed by several hours and created reports that are organized into cover pages, indexes and other variables that properly classify the assignment. 

The new software inserts fields directly into the output file with a defined business logic, eliminating the need to fill in information like names, case numbers, and agencies. Ultimately, the new Transcriptions Pages are more accurate and secure, cutting down work hours while increasing accuracy of their work.

Processing speech to text files was only the beginning; the other part that needed to be addressed was the user interface for managers and administrators. All tasks after MP3 conversions are communicated inside the interface web application where users log in. What used to be manual inputs of communications turned into automatic reminders and pop-ups for managers and QA administrators. There was no longer a need for typed notes to identify and list what corrections were made, who the file belonged to, or when the tasks were completed. Those items are now automatically generated.

Administrators can log in to see progress and completed work in an organized dashboard view. Project Managers can see status reports and real time data to make quick, operational decisions. A User Management feature allows Executive Leadership to enable and disable users when key personal choices are made on assignments. These are all additional features that didn’t exist, or didn’t efficiently function, and lead to cutting down 75% of processing time.

Customizing their workflow of audio-to-text conversions, quality transcriptions, manager and user interfaces, and automated updates and secured files, allows Conduit to provide speedy, accurate, and secure work with a higher guarantee to their clients. 

For Conduit, it was all about simplifying the process. Endertech software developers examined each stage of their workflow and connected it to the next. The new custom software allows for greater flexibility and solves problems in finer detail. Thanks to this upgrade, Conduit Transcriptions can now maintain an efficient flow of high-quality work, and explore new opportunities in the future.