Skip to main content

​
Overview

The Transcripts API provides access to complete earnings call transcripts stored in S3. Metadata is stored in PostgreSQL for fast querying, while full transcript text is retrieved from cloud storage.

​
Get Transcript

Retrieve a complete earnings call transcript. Endpoint: GET /transcript/{ticker}/{year}/{quarter} Authentication: Optional (works without authentication)
cURL
curl https://api.financeagent.com/transcript/TSLA/2024/4

​
Path Parameters

​
ticker
string
required
Company ticker symbol (e.g., “TSLA”, “AAPL”)
​
year
integer
required
Fiscal year (e.g., 2024)
​
quarter
integer
required
Fiscal quarter (1, 2, 3, or 4)

​
Response

​
success
boolean
Whether the transcript was found
​
ticker
string
Company ticker symbol
​
company_name
string
Company name
​
year
integer
Fiscal year
​
quarter
integer
Fiscal quarter
​
date
string
Earnings call date
​
transcript_text
string
Complete transcript text
​
transcript_length
integer
Length of transcript in characters
​
source
string
Data source (always “bucket” for S3)
​
metadata
object
Additional metadata about the transcript

​
Example Response

{
  "success": true,
  "ticker": "TSLA",
  "company_name": "Tesla Inc.",
  "year": 2024,
  "quarter": 4,
  "date": "2024-01-24",
  "transcript_text": "Operator: Good day, and thank you for standing by...\n\nElon Musk: Thanks for joining us today...",
  "transcript_length": 45232,
  "source": "bucket",
  "metadata": {
    "duration": "60 minutes",
    "participants": ["Elon Musk", "Zachary Kirkhorn"]
  }
}

​
Get Transcript with Highlights

Retrieve a transcript with specific sections highlighted based on relevant chunks. Endpoint: POST /transcript/with-highlights Authentication: Optional 
curl -X POST https://api.financeagent.com/transcript/with-highlights \
  -H "Content-Type: application/json" \
  -d '{
    "ticker": "AAPL",
    "year": 2024,
    "quarter": 3,
    "relevant_chunks": [
      {
        "chunk_id": "chunk-42",
        "char_offset": 15200,
        "chunk_length": 800,
        "chunk_text": "Our Services revenue grew 14% year over year..."
      }
    ]
  }'

​
Request Body

​
ticker
string
required
Company ticker symbol
​
year
integer
required
Fiscal year
​
quarter
integer
required
Fiscal quarter (1-4)
​
relevant_chunks
array
required
Array of relevant chunks to highlight

​
Response

​
success
boolean
Request status
​
transcript_text
string
Original transcript text
​
highlighted_transcript
string
Transcript with HTML <mark> tags around highlighted sections
​
transcript_length
integer
Length of transcript in characters
​
metadata
object
Transcript metadata

​
Highlighted Output Example

Operator: Good day, and thank you for standing by...

<mark class="highlighted-chunk" data-chunk-id="chunk-42">
Our Services revenue grew 14% year over year to reach a new all-time high of $22.3 billion. This growth was driven by strong performance across our services portfolio.
</mark>

We also saw continued momentum in our Products segment...

​
Highlighting Logic

The API uses a two-step approach to highlight relevant sections:
  1. Primary Method: Uses char_offset and chunk_length for precise positioning
  2. Fallback Method: Performs text search on chunk_text if offset is unavailable

​
Minimum Highlight Length

Highlighted regions are expanded to a minimum of 1,200 characters with a 400-character look-ahead to provide context.

​
Error Responses

​
Transcript Not Found

{
  "detail": "Full earnings transcript not yet available for TSLA 2025 Q1"
}
Status Code: 404 Not Found

​
Storage Error

{
  "detail": "Could not load transcript from storage"
}
Status Code: 503 Service Unavailable

​
Use Cases

​
1. Display Full Transcript

# Fetch and display a complete transcript
response = requests.get(
    "https://api.financeagent.com/transcript/MSFT/2024/2"
)
transcript = response.json()
print(f"Transcript for {transcript['company_name']} Q{transcript['quarter']} {transcript['year']}")
print(transcript['transcript_text'])

​
2. Highlight Search Results

# After getting search results from chat API, highlight matching sections
relevant_chunks = [
    {
        "chunk_id": "chunk-15",
        "char_offset": 8500,
        "chunk_length": 650,
        "chunk_text": "Cloud revenue increased by 23%..."
    }
]

response = requests.post(
    "https://api.financeagent.com/transcript/with-highlights",
    json={
        "ticker": "MSFT",
        "year": 2024,
        "quarter": 2,
        "relevant_chunks": relevant_chunks
    }
)

html_content = response.json()['highlighted_transcript']
# Render html_content in your application

​
3. Compare Multiple Transcripts

# Fetch transcripts for trend analysis
quarters = [1, 2, 3, 4]
transcripts = []

for q in quarters:
    response = requests.get(
        f"https://api.financeagent.com/transcript/NVDA/2024/{q}"
    )
    transcripts.append(response.json())

# Analyze or display side-by-side

​
Caching

Transcript text is cached in-memory (max 50 transcripts) to reduce S3 API calls and improve response times. Metadata queries always hit PostgreSQL for accuracy.

​
Performance

  • Metadata Query: ~50-100ms (PostgreSQL)
  • Transcript Fetch: ~200-500ms (S3, first request)
  • Cached Transcript: ~10-20ms (in-memory)
  • Highlighting: ~50-100ms (text processing)
For best performance, use char_offset in highlight requests rather than relying on text search fallback.

Build docs developers (and LLMs) love

Get started for freeTalk to us