Overview
The NFL Strike Word dataset contains 3,894 game transcripts spanning from January 2, 1966 to January 4, 2026. Seeing as there are over 250 NFL games per season, and the league just turned 60 years old, this dataset covers roughly 1/4th of all NFL games. Each transcript represents a complete game broadcast, including play-by-play commentary, analysis, and other spoken content from NFL games. It excludes audio from advertisements. Note, however, it neither excludes or demarcates audio from other speakers, such as a sideline reporter or words from mic’d up players.Distribution by Team
The dataset includes transcripts from games involving all present and some past NFL teams. The distribution below shows the total number of games each team appears in. You’ll notice that recently relocated teams like the Las Vegas Raiders and Los Angeles Chargers have the least data while the most permanent franchises like the Philadelphia Eagles, Dallas Cowboys, and Miami Dolphins have the most data.
Note that we abbreviate each time with up to three letters. These team abbreviations are the only representation of a team accepted by the
home_team and away_team parameters of our endpoint.Distribution by Season
This dataset spans 58 NFL seasons (we couldn’t find any games from 1966 or 1967). Unfortunately, transcripts for recent seasons is the most difficult to come by.
Distribution by Week
The dataset includes regular season games (organized by week) and playoff games (organized by round). Regular season games show the week number (Week 1-18), while playoff games show the round: Wild Card (WC), Divisional (DIV), Conference Championship (CON), and Super Bowl (SB).