Conference Call Transcripts

Real-Time Conference Call Transcripts for Every Public Company in the US and Canada

Earnings calls, investor days, and events - transcribed and delivered in real time. 

Here's how we built it and why it matters for retail investors. 

- View the call summary

- Read the entire speaker diarized transcript

- Ask questions against the transcript

- Enriched participant information (company, title)

The Gap

If you're a retail investor, you've probably felt this pain: a company you own reports earnings, and by the time you find a transcript, the stock has already moved 8%. Institutional desks have real time audio feeds, dedicated analysts, and transcription pipelines. Everyone else gets the leftovers hours later.

We built Benzinga's transcript product to close that gap. Every conference call across US and Canadian public markets, including earnings calls, analyst days, guidance revisions, and special events, transcribed live and available the moment words are spoken.

First Party Data

Most financial data platforms are redistributors. They license data from upstream providers, repackage it, and pass it along. They're always second in line.

We wanted to be the source. We built our own transcription infrastructure, making Benzinga a first party data provider. We're not waiting for a third party to process a call and hand us the output. We're capturing and transcribing it ourselves, in real time. That makes us one of the first places the information exists in structured form.

Being a primary source rather than a redistributor is the only way to truly deliver on our core mission: giving retail investors the same access to market information that institutional players have always had. If you're relying on someone else's pipeline, you're always a step behind, and so are your users.

Coverage

We cover essentially every public company that hosts a conference call in the US and Canada, across every type of corporate event that moves markets: earnings calls, investor days, guidance updates, M&A announcements, analyst days, special events, and shareholder meetings. 

Each transcript is enriched with structured metadata including speaker identification, company ticker, event type, and timestamps, all delivered through the same APIs that power the rest of the Benzinga platform. 

Why Go as the Language

Peak earnings season is a concurrency problem. Hundreds of companies report on the same handful of days each quarter. When the first week of February hits, you might see 200+ simultaneous conference calls, each requiring a dedicated audio stream, real time transcription pipeline, and output delivery channel.

We wrote the entire transcription backend in Go. Go's goroutine model gives us lightweight concurrency without the overhead of OS threads. Each incoming call spins up its own pipeline: audio ingestion, speech to text processing, transcript assembly, and API delivery, all as goroutines communicating through channels.

During peak earnings season, the system scales from a handful of calls to a few hundred simultaneous sessions without rewriting the architecture. (We made some mistakes early on with pod scaling. That’s a story for another day!) 

func (s *Service) RealTimeTranscriber(ctx context.Context, s3 s3.Storage, uploadChan chan *domain.LocalProcess, producer *publish.Producer) {
	// Process each HLS URL in a separate goroutine
	for {
		select {
		case call := <-s.RealTimeChan:
			go s.handleStream(ctx, call, s3, uploadChan, producer)
		case call := <-s.NonLiveChan:
			go s.PostProcessor(ctx, &domain.LocalProcess{
				Req:         call,
				File:        call.CallURL,
				FileType:    domain.URLFileType,
				Meta:        nil,
				PostProcess: false,
			})
		case call := <-uploadChan:
			go s.PostProcessor(ctx, call)
		default:
			time.Sleep(time.Millisecond * 100)
		}
	}
}

Lean Team

Incumbents in this space have raised tens of millions of dollars to build what we've shipped with a small, focused team. We're not trying to out-hire or out-spend anyone. We're building lean infrastructure that punches above its weight by choosing the right tools, keeping the architecture simple, and staying close to the problem. 

API First

Transcripts are available through the Benzinga API as structured data. Developers can pull them into trading algorithms, feed them into language models, or build custom alerting on keywords and sentiment. The same data a fundamental analyst reads is available programmatically for quantitative workflows.

Every transcript comes with AI generated highlights out of the box: key takeaways, management tone, and forward guidance extracted and summarized so you can get the signal without sitting through an hour long call. 

The Whole Point

The thesis behind Benzinga has always been simple: retail investors deserve the same data that institutional desks get. Real time transcripts are one of the last major gaps. When a CEO makes an off script comment about supply chain issues in a Q&A, everyone should have access to that information at the same time, not just the funds with a Bloomberg Terminal.

That's what we're building - let me know if you have any feedback or suggestions on how we can keep building out this product!