Speaker-labelled transcripts
Each turn is attributed to the real person with the role they held at the time. Unverified voices show as a neutral speaker number rather than a guessed name.
Get earnings call transcripts as JSON, with speaker labels and positions in the call audio. Query by ticker and fiscal quarter in your own code, or read the same calls in ChatGPT or Claude through the MCP server.
Using ChatGPT or Claude? Add this data to your AI — no code needed.
Free: 100 requests a day, no credit card. One account works with REST and MCP.
Both surfaces read the same store the Equibles pages render, so a figure on the site and a figure in your code agree.
Each turn is attributed to the real person with the role they held at the time. Unverified voices show as a neutral speaker number rather than a guessed name.
One record groups a quarter's call artefacts, audio, transcript, slide deck and the 8-K earnings release, with the guidance extracted from that release when a verifier-approved extraction backs it.
A verifier-approved brief and insight scores for each call, plus tone and themes, so a model can start from the summary and drill into the exact turn.
Every endpoint below is a GET under the api.equibles.com/v1 base URL, authenticated with an API key. Each MCP tool answers the same question in plain English from Claude, ChatGPT or any connected client.
GET /v1/stocks/{ticker}/earnings-calls/{fiscalYear}/{fiscalQuarter}GET /v1/stocks/{ticker}/earnings-calls/{fiscalYear}/{fiscalQuarter}/speakersGET /v1/stocks/{ticker}/earnings-briefsGET /v1/stocks/{ticker}/call-insightsGET /v1/stocks/{ticker}/investor-eventsOne free key covers the MCP server and the REST API; the browser needs no key at all.
curl "https://api.equibles.com/v1/stocks/NVDA/earnings-calls/2027/1/speakers?limit=200" -H "Authorization: Bearer eq_YOUR_KEY" curl "https://api.equibles.com/v1/stocks/NVDA/earnings-calls/2027/2/speakers?limit=200" -H "Authorization: Bearer eq_YOUR_KEY"
“Using Equibles, compare forward revenue guidance in NVIDIA's FY2027 Q1 and Q2 earnings calls. State which fiscal quarter each target covers and cite the transcript passages.”
Name Equibles in the prompt so the model calls the tool instead of answering from memory. Connection guides
Read on 30 September 2026. Two calls, with the period being guided kept separate from the call's fiscal quarter.
| Call | Call date | Revenue guidance applies to | Revenue target |
|---|---|---|---|
| FY2027 Q1 | 20 May 2026 | FY2027 Q2 | $91 billion, ±2% |
| FY2027 Q2 | 26 August 2026 | FY2027 Q3 | $108 billion, ±2% |
The target for the following quarter rose by $17 billion (18.7%). These targets cover different quarters: this is a comparison of successive outlooks, not a revision to guidance for the same period. Fetch both speaker-turn responses and follow hasMore/offset if a longer call needs another page.
FY2027 Q1 transcript: Let me turn to the outlook for the second quarter. Total revenue is expected to be $91 billion, plus or minus 2%.
FY2027 Q2 transcript: Let me turn to the outlook for the third quarter. Total revenue is expected to be $108 billion, plus and minus 2%.
Fiscal year and quarter follow the company's own calendar, not the calendar year: NVIDIA's May 2026 call is FY2027 Q1. On the MCP tool, omit both to get the latest call.
An empty guidance list is readable through its status: approved means the extraction ran and the release genuinely guided nothing more, notExtracted means nobody has extracted the release yet, and null means no 8-K earnings release is linked to the event.
Source: Company earnings webcasts, with the 8-K releases from SEC EDGAR.
Choose the interface that fits the work; all three read the same store.
It returns the text of a company's earnings call as structured data: the speaker turns in order, who said each one and in which role, plus the event record, the 8-K release and the guidance extracted from it, so code and AI models can quote the call rather than a summary of it.
Each turn is attributed to the real executive or analyst with the role they held at the time. Identities appear only when the resolution is trusted, either auto-resolved or human-reviewed; unverified voices show as a neutral speaker number.
The company's own fiscal calendar. NVIDIA's May 2026 call is FY2027 Q1. On the MCP tool you can omit the year and quarter to get the latest call; on the REST API both are path segments.
Use the quarter's speakers endpoint. Start at offset 0, request up to 200 turns per page, and continue while hasMore is true. Increase offset by turnCount after each response. Keep speaker labels, audio positions and any corrections with the text.
No. Coverage varies by company and event. Read the call event's availability flags before requesting its transcript; the event record, captured transcript, audio and slides are separate artefacts.
Yes. Connect the Equibles MCP server and ask for a company's call by ticker and quarter; the GetEarningsCallTranscript tool returns the speaker-labelled turns and pages through long calls. The REST API returns the same data as JSON.
Yes, over MCP. Pass MIC:TICKER, such as XPAR:AIR for Airbus in Paris, to GetEarningsCallEvent or GetEarningsCallTranscript to read a European, Chinese, Hong Kong, Japanese or Canadian listing's calls where captured, each transcript in the language the call was held in. The REST endpoints take US tickers.
Yes. The Equibles free tier includes 100 requests a day, shared across the REST API and MCP server, with no credit card. Paid plans raise the limit.
One free key for transcripts, call events, guidance and briefs, over REST and MCP.