sleeping-ai/SUNO-XX1

SUNO XX1

You can find the raw database here

Introducing SUNO XX1, a dataset we constructed after community request to receive the entire SUNO load and having a better and comprehensive 1:1 mirror database of the SUNO site. It is an expansion on-top of SUNO-1M we released earlier.

We are releasing estimated 2M songs and a 1:1 mirror of SUNO database.

this repo will be constantly uploaded over the course of next 2 weeks.

LICENCE

We release this under CC-by-nc-nd 4.0, preventing people from below acts:

1. Sharing the dataset from personal storage
2. We don't allow any commercial use + any derivative of this dataset + any re-upload attempt.

Please, follow this rules for making sure we can release more research datasets legally and push state-of-art research.

Data Field Descriptions

• id: Unique identifier for the song entry.
• user_id: Unique identifier for the user who created the song.
• audio_url: URL to the song's audio file.
• video_url: URL to the song's video file.
• allow_comments: Boolean indicating if comments are allowed.
• artist_clip_id: Identifier for the artist's clip, if applicable.
• avatar_image_url: URL to the user's avatar image.
• can_publish_with_vocal: Boolean indicating if publishing with vocals is allowed.
• can_remix: Boolean indicating if the song can be remixed.
• caption: Optional caption for the song.
• comment_count: Number of comments on the song.
• concat_history: History of concatenated clips, if any.
• continue_at: Timestamp for resuming song creation.
• cover_clip_id: Identifier for the cover clip, if applicable.
• created_at: Timestamp of when the song was created.
• display_name: User's display name.
• display_tags: Tags displayed for the song.
• duration: Duration of the song in seconds.
• edit_session_id: Identifier for the editing session.
• edited_clip_id: Identifier for the edited clip, if applicable.
• entity_type: Type of entity (e.g., song_schema).
• explicit: Boolean indicating if the song contains explicit content.
• flag_count: Number of flags reported on the song.
• from_index: Starting index for pagination, if applicable.
• handle: User's unique handle.
• has_vocal: Boolean indicating if the song includes vocals.
• image_large_url: URL to the large version of the song's image.
• image_s3_id: S3 identifier for the song's image.
• image_url: URL to the song's image.
• include_future_s: Future settings inclusion flag.
• include_history_s: History settings inclusion flag.
• infill: Infill data for song generation.
• infill_context_end_s: End time for infill context.
• infill_context_start_s: Start time for infill context.
• infill_dur_s: Duration of infill segment.
• infill_end_s: End time for infill segment.
• infill_start_s: Start time for infill segment.
• is_contest_clip: Boolean indicating if the song is part of a contest.
• is_handle_updated: Boolean indicating if the handle has been updated.
• is_liked: Boolean indicating if the song is liked by the user.
• is_owned: Boolean indicating if the song is owned by the user.
• is_public: Boolean indicating if the song is publicly accessible.
• is_remix: Boolean indicating if the song is a remix.
• is_trashed: Boolean indicating if the song is marked as trashed.
• major_model_version: Major version of the model used to generate the song.
• metadata: Additional metadata about the song (e.g., tags, prompt).
• model_name: Name of the model used for song generation.
• name: Optional name of the song.
• negative_tags: Tags indicating undesirable attributes.
• page_size: Size of pagination page, if applicable.
• persona: Persona associated with the song.
• persona_id: Identifier for the persona.
• play_count: Number of times the song has been played.
• preview_url: URL to the song's preview.
• priority: Priority level for song generation.
• prompt: Text prompt used for song generation.
• public_songhot: Indicator of song popularity.
• refund_credits: Boolean indicating if credits were refunded.
• result: Result data from song generation.
• root_clip_id: Identifier for the root clip, if applicable.
• source: Source of the song.
• status: Current status of the song (e.g., complete).
• stream: Boolean indicating if the song is streamable.
• tags: Tags associated with the song.
• task: Task associated with song generation.
• title: Title of the song.
• total_hits: Total number of hits/views for the song.
• type: Type of song or generation.
• upsample_clip_id: Identifier for the upsampled clip, if applicable.
• upvote_count: Number of upvotes the song has received.
• user_display_name: Display name of the user.
• user_handle: Handle of the user.
• user_image_url: URL to the user's image.
• video_cover_url: URL to the video cover image.
• video_upload_height: Height of the uploaded video.
• video_upload_width: Width of the uploaded video.