Sign in with Google or email to join.
Every word arrives within seconds of being spoken. Streaming transcript, no replay delay.
Each line tagged to the speaker who said it, named as the call unfolds and corrected when needed.
On this call: Dave Fildes, Andy Jassy, Brian Olsavsky · 12 covering analysts
Every word attributed to the speaker who said it. Streamed within seconds of being spoken. This is the foundation for every product that operates while the call is live — alerts, signal generation, position sizing, post-call analysis. Build what your competitors can't.
Diarization tells you a speaker changed.
Live Earnings tells you who it was.
Live diarization is not enough. Live Earnings resolves the speaker.
Surface guidance, surprises, analyst pressure points in the seat — sub-5-second from spoken word to your terminal.
Replace transcription vendors that ship hours later. Live attributed copy, ready to highlight, quote, embed.
Feed attributed turns into your own models for sentiment, deviation-from-guidance, Q&A pattern detection.
pip install live-earningsfrom live_earnings import LiveEarningsClient, Transcript
client = LiveEarningsClient(token=...)
transcript = Transcript()
for env in client.stream("nvda-q1-fy27"):
if turn := transcript.apply(env):
print(turn.name or "…", turn.text or "")npm i live-earnings-sdkimport { LiveEarningsClient, Transcript } from "live-earnings-sdk";
const client = new LiveEarningsClient({ token });
const transcript = new Transcript();
for await (const env of client.stream("nvda-q1-fy27")) {
const turn = transcript.apply(env);
if (turn) console.log(turn.name ?? "…", turn.text ?? "");
}