diff options
Diffstat (limited to 'code/ai_recommendations.py')
| -rw-r--r-- | code/ai_recommendations.py | 36 |
1 files changed, 19 insertions, 17 deletions
diff --git a/code/ai_recommendations.py b/code/ai_recommendations.py index 9c0831f..1fd9c3f 100644 --- a/code/ai_recommendations.py +++ b/code/ai_recommendations.py @@ -5,7 +5,9 @@ from config import OPENAI_API_KEY import openai -async def add_song_recommendations(openai_client, bot_user, player, number, inputs, retries: int = 1): +async def add_song_recommendations( + openai_client, bot_user, player, number, inputs, retries: int = 1 +): input_list = [f'"{song} by {artist}"' for song, artist in inputs.items()] completion = ( @@ -14,25 +16,25 @@ async def add_song_recommendations(openai_client, bot_user, player, number, inpu { "role": "user", "content": f""" - BACKGROUND: You're an AI music recommendation system with a knack for understanding - user preferences based on provided input. Your task is to generate a list - of {number} songs that the user might enjoy, derived from a given list of {number} songs. - The input will be in the format of - ["Song-1-Name by Song-1-Artist", "Song-2-Name by Song-2-Artist", ...] - and you need to return a list formatted in the same way. + BACKGROUND: You're an AI music recommendation system with a knack for understanding + user preferences based on provided input. Your task is to generate a list + of {number} songs that the user might enjoy, derived from a given list of {number} songs. + The input will be in the format of + ["Song-1-Name by Song-1-Artist", "Song-2-Name by Song-2-Artist", ...] + and you need to return a list formatted in the same way. - When recommending songs, consider the genre, tempo, and mood of the input - songs to suggest similar ones that align with the user's tastes. Also, it - is important to mix up the artists, don't only give the same artists that - are already in the queue. If you cannot find {number} songs that match the - criteria or encounter any issues, return the list ["NOTHING FOUND"]. + When recommending songs, consider the genre, tempo, and mood of the input + songs to suggest similar ones that align with the user's tastes. Also, it + is important to mix up the artists, don't only give the same artists that + are already in the queue. If you cannot find {number} songs that match the + criteria or encounter any issues, return the list ["NOTHING FOUND"]. - Please be sure to also only use characters A-Z, a-z, 0-9, and spaces in the - song and artist names. Do not include escape/special characters, emojis, or - quotes in the output. + Please be sure to also only use characters A-Z, a-z, 0-9, and spaces in the + song and artist names. Do not include escape/special characters, emojis, or + quotes in the output. - INPUT: {input_list} - """, + INPUT: {input_list} + """, } ], model="gpt-3.5-turbo", |