Digital Nirvana’s Transcription Solutions for Financial Markets

Industry Challenges

Notably, financial analysts rely on these transcripts to build their economic models and create research reports that get distributed to their clients, who in turn use the information to make investment decisions. If one number goes wrong or one decimal is out of place in the transcript, it can mean hundreds of millions of dollars get moved to the wrong spot, and legal liabilities come into play.

With so much riding on them, the transcripts must be precise and detailed. So not just any transcription solution will do. Not only that, but the sheer volume of transcription — earnings calls alone can total at least 500 quick-turn transcripts per day at the height of the quarter — requires scalable infrastructure and specialized transcriptionists who understand the terminology and can handle the volume without sacrificing accuracy, detail, and budget. Few transcription providers are up to the task.

The new transcription solution had to address the following criteria:

• Seasonal volume adjustment: the ability to scale up and down as seasonal variations dictate.
• Infrastructure: the technology and processes to record calls reliably and the scalability to record hundreds of calls at the same time.
• Quality: accurate transcripts across all seasonal volumes.
• TAT: maintain a streaming turnaround time (TAT) of less than two minutes and final TAT of three hours, even during peak periods.
• Cost : keep the project cost within the allocated budget.
• Flexibility: the ability for the client to change, cancel, or add calls during the day as circumstances change.
• Communication: seamless communication with the transcription service provider, who would literally work as an extension of the financial organization’s team.

Solution

Digital Nirvana began providing earnings call transcripts in 2001. In the years since, we have developed — and continually refine — the technology and processes to handle all the variables and nuances of financial transcription.

Digital Nirvana offers transcription through two AI-based avenues:

• An automated SaaS solution called Trance that lets you manage the process yourself.
• Full-service, turnkey transcription managed by our specialized in-house team.

You choose what's best for your business.

Trance: Build efficient, accurate transcription into your existing workflow

Leading companies in the financial space use Trance to stream real-time transcripts of earnings and corporate conference calls conducted by Fortune 500 companies. Key attributes:

Trance unites cutting-edge speech-to-text (STT) technology and other AI-driven processes with cloud-based architecture to reduce the time and cost of delivering accurate transcripts.

In an industry first, Trance automatically identifies music, silence, and other nontext audio. It provides intelligent line breaks, preventing the awkward splitting of words or dangling punctuation at the end of a line. Trance also keeps multiword proper names together. Robust presets and grammar rules ensure that lines do not start with misplaced punctuation, such as the apostrophe in a contraction. 

Trance automatically detects speaker-change, such as when the conversation goes from the CEO to the operator to the CFO. The algorithm automatically inserts line breaks to account for the change in voices.

An advanced, AI-driven STT engine trained in the language of finance makes it possible to generate highly accurate, industry-specific content in real time. Through machine learning, this automated speech recognition engine will only improve over time.

The service scales up to 400 hours of live transcription in a day during the earnings-release season and scale down to fewer hours during other days in a calendar quarter.

Outsource your transcription to Digital Nirvana

Many financial organizations choose Digital Nirvana to handle the real-time transcription function from beginning to end, which allows them to eliminate dedicated teams working on transcription.

Digital Nirvana’s finance operations team builds and maintains a calendar of earnings calls for as many as 4,500 of the most followed public companies by the investors and competitors. This effort alone is of critical importance in helping users track which companies are reporting at which times and how to access the call (phone number, webcast login, etc.).

Once contracted to cover a call, the team uses an internal version of Trance to generate real-time transcripts. Digital Nirvana relies on a stable pool of specialized transcriptionists who understand the numbers and terminology to review the transcripts and ensure their accuracy.

At the height of the earnings-reporting period each quarter, Digital Nirvana’s transcription infrastructure scales up to handle at least 500 earnings calls per day, recording those calls, using speech-to-text to transcribe them, and then having a human edit each one — and turning it all around in three to six hours to ensure timely reporting that keeps up with the markets and investment cycles.

Because this financial organization preferred not to live-stream direct speech-to-text output to analysts, Digital Nirvana worked with the company to develop what we call a near-live transcript. It works like this:

For the first version of the transcript, a speech-to-text engine generates a transcript in real time. Digital Nirvana’s highly trained financial transcriptionists quickly review the engine’s output to correct blatant errors, insert paragraph breaks and speaker names, and generally make the transcript more readable.

Then we stream that version to the client’s chosen destination for viewing by its customers, usually within 30 seconds but up to three minutes.

The next version of the product, akin to a raw transcript that has not yet been fully edited, goes up within half an hour of the call.

From there, the Digital Nirvana team combs through the transcript in detail to correct errors, thoroughly review the numbers, add speaker tags, and apply the company’s style guide, branding, and formatting; with that done, the final edited transcript is shipped to the client’s database within six hours.

Differentiators:

This financial data provider summarized the benefits of working with Digital Nirvana this way:

1. Trust — The same earnings call can be transcribed once and sold to multiple financial organizations, but for this company, that practice presents an IP issue. They had been burned by other transcription providers that promised to do an exclusive transcript but clearly sold them a copy. Digital Nirvana proved that when we cover the same call for multiple companies, we have various teams in place making numerous transcripts of the same call. That exclusivity garnered trust.

2. Quality — Maintaining consistent quality is a challenge for many transcription providers. It is easy to do it when volumes are low, but when volumes are high on a given day, and turnaround time doesn’t change, then quality and accuracy can deteriorate. Digital Nirvana delivers financial transcripts with the highest possible accuracy, as evidenced by its Q Score methodology. Typical quality checks look for errors in grammar or spelling, but in financial transcription, an errant number has a much more significant impact than a misspelled word. So Digital Nirvana came up with a weighted average quality score that makes it possible to quantify the quality of a financial transcript. Now the Q Score has become an industry standard for measuring transcript quality.

3. Quick and Efficient — Because of the technology and operational efficiencies, Digital Nirvana can turn complete transcripts around very quickly — usually within three to four hours compared to six hours or more with other providers.

Today, Digital Nirvana is by far this financial data company’s largest transcription provider.

Because this significant financial data provider is also a broadcast media outlet with a TV, radio, online, and podcast presence, there is potential to use Digital Nirvana’s transcription solutions to handle so much more than earnings calls.

Benefits

Speech-to-text technology and other AI-driven processes radically reduce the time and cost of delivering accurate transcripts.

Generates fast and automatic transcripts.

Provides AI-driven STT with expertise in the language and terminology of finance.

Makes transcripts more readable with AI-driven capabilities.

Delivers up to 400 hours of live transcription per day at peak season.

Finance operations team tracks a seasonal calendar of earnings calls.

With the managed service, specialized transcriptionists review transcriptions and correct errors.