version: 2 models: - name: dim_clip description: | Dimension table containing comprehensive metadata and attributes for all audio clips generated or uploaded to the Suno platform. This table serves as the central source of truth for clip information including creation details, user interactions, content metadata, generation parameters, and status information. This is a view that adds enhanced categorization and metadata on top of the intermediate int_dim_clip model. This model provides cleaned and enriched clip data with additional computed fields like create_mode_cleaned, model_category, category, history_length, is_made_in_edit, and has_augmented_tags. config: schema: PROD group: clips meta: dagster: automation_condition: hourly_cron_with_eager_historical_backfill_condition partitions_def: type: hourly start_date: 2023-05-12 end_offset: 0 backfill_policy: max_partitions_per_run: -1 # full backfill columns: - name: clip_id description: "Unique identifier for the clip (primary key)" - name: p_date description: "Partition date - date when the clip was created (used for table clustering)" - name: p_hour description: "Partition hour - hour when the clip was created (used for table clustering)" - name: created_at description: "Timestamp when the clip was created (with timezone information)" - name: user_id description: "ID of the user who created the clip" - name: request_id description: "Unique identifier for the generation request that created this clip" - name: s3_id description: "S3 object identifier for the clip's audio file" - name: platform description: "Platform where the clip was created (web, mobile, etc.)" - name: display_index description: "Index/order of the clip within a batch generation request" - name: initial_title description: "Original title assigned to the clip at creation" - name: time_used_sec description: "Time in seconds taken to generate the clip" - name: type description: "Type of clip generation (e.g., song, instrumental, remix)" - name: task description: "Specific task or mode used for clip generation" - name: priority description: "Priority level assigned to the generation request" - name: duration_sec description: "Duration of the generated clip in seconds" - name: is_refund_credits description: "Boolean indicating if credits were refunded for this clip" - name: is_bot description: "Boolean indicating if the clip was generated by a bot user" - name: credit_cost description: "Number of credits consumed to generate this clip" - name: lyrics_at_gen description: "Lyrics text that was used during clip generation" - name: simple_prompt description: "Simplified prompt description used for generation" - name: lyrics_final description: "Final lyrics text associated with the clip" - name: is_simple description: "Boolean indicating if this was generated using simple mode" - name: is_make_instrumental description: "Boolean indicating if the clip was generated as instrumental only" - name: style_tags description: "JSON array of style tags applied to the clip" - name: negative_style_tags description: "JSON array of style tags to avoid during generation" - name: request_models description: "JSON array of AI models requested for generation" - name: model_name description: "Name of the AI model actually used for generation" - name: experiment_inf_param description: "JSON object containing experimental inference parameters" - name: experiment_version description: "Version identifier for experimental features used" - name: history description: "JSON object containing generation history and iterations" - name: concat_history description: "Concatenated string representation of generation history" - name: error_type description: "Type of error if clip generation failed" - name: error_message description: "Detailed error message if clip generation failed" - name: upload_has_vocal description: "Boolean indicating if uploaded clip contains vocals" - name: edit_session_id description: "ID of the editing session if this clip was edited" - name: edited_clip_id description: "ID of the original clip if this is an edited version" - name: task_details description: "JSON object containing detailed task-specific information" - name: creation_source description: "Source or method used to create the clip" - name: augmented_tags description: "JSON array of additional tags added to the clip" - name: free_quota_category description: "Category classification for free quota usage" - name: lyrics_language description: "Language of the lyrics in the clip" - name: simple_lyrics_model description: "Model used for simple lyrics generation" - name: created_session_id description: "ID of the user session when the clip was created" - name: clip_uid description: "Unique user-facing identifier for the clip" - name: is_audio_watermark description: "Boolean indicating if audio watermarking was requested (default: false)" - name: is_audio_watermarked description: "Boolean indicating if the clip has audio watermarking applied (default: false)" - name: user_tier description: "Subscription tier of the user who created the clip" - name: is_deleted description: "Boolean indicating if the clip has been soft deleted" - name: create_mode description: "Mode or method used to create the clip" - name: combined_from_clip_ids description: "JSON array of clip IDs that were combined to create this clip (if applicable)" - name: create_mode_cleaned description: "Cleaned creation mode (simple, custom, or from create_mode field)" - name: model_category description: "Categorized model version (v3.5, v4, v4.5, v4.5+, v5, stem, older_model, no_model)" - name: category description: "Detailed clip category based on type and task (normal_gen, cover_gen, extend_gen, etc.)" - name: history_length description: "Number of steps in the creation history" - name: is_made_in_edit description: "Boolean indicating if the clip was created in edit mode" - name: has_augmented_tags description: "Boolean indicating if the clip has augmented tags" - name: metadata description: "Parsed JSON object containing all clip metadata from the raw metadata_json field" - name: web_client_pathname description: "Web client pathname from metadata, indicating the page/route where the clip was created" - name: is_public description: "Boolean indicating if the clip is public (from rds_public_clip table)" - name: clip_initial_metrics description: | View that provides initial interaction and satisfaction metrics on user-owned clips within a 120-minute window after creation. This is a view over the int_clip_initial_metrics intermediate model, which contains the actual incremental logic. Tracks plays, likes, dislikes, shares, downloads, and privacy toggles by the clip owner to measure early engagement and satisfaction. Includes both direct satisfaction (explicit user actions) and indirect satisfaction (play behavior) indicators. config: schema: PROD group: clips meta: dagster: automation_condition: hourly_cron_with_eager_historical_backfill_condition partitions_def: type: hourly start_date: 2023-05-12 end_offset: -3 backfill_policy: max_partitions_per_run: -1 asset_key: ["clip_initial_metrics"] columns: - name: clip_id description: "Unique identifier for the clip (primary key)" - name: user_id description: "User ID who created the clip" - name: user_uid description: "User UID who created the clip" - name: clip_created_at description: "Timestamp when the clip was created" - name: window_minutes description: "Time window in minutes for measuring initial metrics (120 minutes)" - name: window_start_time description: "Start of the measurement window (truncated to hour of clip creation)" - name: window_end_time description: "End of the measurement window (window_start_time + window_minutes)" - name: play_cnt description: "Total number of plays by the clip owner within the time window" - name: play_duration_seconds description: "Total play duration in seconds by the clip owner within the time window" - name: n_like_events description: "Number of ClipLike events by the clip owner within the time window" - name: n_undolike_events description: "Number of ClipUndoLike events by the clip owner within the time window" - name: net_like_events description: "Net like events (n_like_events - n_undolike_events)" - name: n_dislike_events description: "Number of ClipDislike events by the clip owner within the time window" - name: n_undodislike_events description: "Number of ClipUndoDislike events by the clip owner within the time window" - name: net_dislike_events description: "Net dislike events (n_dislike_events - n_undodislike_events)" - name: n_share_events description: "Number of ClipShare events by the clip owner within the time window" - name: n_download_events description: "Number of ClipDownload events by the clip owner within the time window" - name: n_makepublic_events description: "Number of ClipMakePublic events by the clip owner within the time window" - name: n_makeprivate_events description: "Number of ClipMakePrivate events by the clip owner within the time window" - name: net_public_toggle_events description: "Net public toggle events (n_makepublic_events - n_makeprivate_events)" - name: net_positive_interactions description: | Combined positive interaction score (net_like_events - net_dislike_events + share_events + download_events + net_public_toggle_events) - name: is_satisfactory_indirect description: | Indirect satisfaction indicator based on play metrics (VARIANT) (play_cnt >= 3 AND play_duration_seconds >= 60 AND no net dislikes) - name: is_satisfactory_direct description: | Direct satisfaction indicator based on explicit user actions (VARIANT) (positive interactions without net dislikes) - name: is_satisfactory description: | Overall satisfaction indicator (VARIANT) (is_satisfactory_indirect OR is_satisfactory_direct) - name: metadata description: "Additional metadata including model_name and other clip attributes (VARIANT)" - name: p_date description: "Partition date (date of clip creation)" - name: p_hour description: "Partition hour (hour of clip creation)" - name: agg_play_info_hourly_v2 description: Hourly aggregated play information on clips, from fact_play. The table is indexed by user ID/UID, clip ID, date + hour, and play duration threshold. config: schema: PROD on_schema_change: append_new_columns group: listening freshness_policy: type: time_window meta: dagster: partitions_def: type: hourly start_date: 2024-06-01 end_offset: -1 partition_mappings: - asset_key: fact_play type: identity - asset_key: dim_user type: identity automation_condition: hourly_cron_with_eager_historical_backfill_condition backfill_policy: max_partitions_per_run: 336 # 2 weeks columns: - name: user_id description: User ID (integer) - name: user_uid description: User UID (UUID) - name: clip_id description: Clip ID - name: is_user_song_owner description: Whether the user is the owner/creator of the song - name: platform description: Platform where the play occurred (web, ios, android, etc). Combines iOS and iPadOS into a single "ios" platform. - name: play_duration_threshold description: Threshold in seconds used to determine valid plays. Currently computing for 0 and 5 second thresholds. - name: play_duration_seconds description: Total duration of plays in seconds in this hour, under the play duration threshold constraint - name: unique_play_duration_seconds description: Total duration of unique plays in seconds in this hour, under the play duration threshold constraint - name: play_count description: Total number of plays in this hour, under the play duration threshold constraint - name: is_high_intent_listening description: JSON object containing breakdown of high intent listening metrics by version. - name: high_intent_listening_breakdown description: JSON object containing breakdown of high intent listening metrics by version. - name: p_date description: Partition date for the hourly aggregation - name: p_hour description: Partition hour (0-23) for the hourly aggregation.