@nodetool-ai/agents 0.8.4-a24.3 → 0.8.5

package/dist/_skills/video-audio-continuity/SKILL.md

@@ -0,0 +1,90 @@
+---
+name: video-audio-continuity
+description: Keep sound continuous across a multi-scene piece cut from generated video — why one clip per scene hard-cuts the audio at every boundary, when to write all the scenes into a single generation instead, the sound-design vocabulary that survives a provider's copyright filter, and how to check the length, the aspect ratio and the mix of what came back. Use when a piece has more than one scene and the video model writes its own audio (Seedance 2, Veo 3, MiniMax H3, Kling 2.6 and later), or when a cut's sound drops out at the shot changes.
+featured: true
+---
+
+# One clip, one bed
+
+A generated clip carries its own audio, and that audio starts and stops at the
+clip's own edges. So the obvious build — one generated clip per shot, laid end
+to end — hard-cuts or drops to silence at every boundary. Stills and contact
+sheets never show it; it only exists on playback, and it is the most common
+audio defect in a multi-scene piece.
+
+## The rule
+
+When sound has to carry across the scene changes, generate **one** clip holding
+**all** the scenes.
+
+- One `generate_video` call, the scenes written as "cut to" beats inside the
+  prompt, and one continuous audio brief across the whole thing.
+- Lay that clip under the sequence as the base bed and time the type and
+  graphics to it.
+- Use a model line that writes native audio, and load its prompting skill first:
+  `seedance-2-prompting` (its JSON beat sheet is built for exactly this — global
+  `audio_note`, one `audio` line per beat), `veo-3-prompting`,
+  `minimax-h3-prompting`, `kling-video-prompting`.
+
+Separate per-shot clips are right in two cases:
+
+- the piece is silent, or
+- the continuity comes from a track you add yourself — narration from
+  `generate_speech` (or `voice_script_lines`, which voices every line of a
+  board's script in one call), a bed from `generate_music` — on its own
+  timeline track. Then the visual beds can be separate and the shots are muted
+  under it. `elevenlabs-audio-prompting` directs the voice and the music plan,
+  `stable-audio-prompting` the bed and any effects; a bed generated with a
+  stated BPM is a bed whose grid `beat-sync-editing` already knows.
+
+Two video lines force the second build: Wan 2.6 takes an audio file as input
+but writes none, and Hailuo has no native audio at all, so a piece on either
+gets its sound from a track of your own.
+
+On a storyboard this means one shot, not seven. A board whose shots each render
+their own native-audio clip cannot be assembled into a continuous mix; decide
+which of the two builds you are doing before the first render, because the fix
+afterwards is a re-render. The board skills (`commercial-beat-sheet`,
+`trailer-template`, `explainer-storyboard`, `music-video-treatment`) each say
+which build they default to; the sound brief itself goes into the shot's
+`motion` and `action`, the only fields the clip prompt is built from.
+
+## A sound brief that is not refused
+
+Providers copyright-filter the generated audio track, and a brief asking for a
+"swell", a "chord", a "score", a "soundtrack" or a named musical genre can come
+back blocked — failing the whole render, not just the audio. For an underscore,
+describe sound design instead:
+
+- Hum, digital pulses, ticking, airy whooshes, risers, sparkle textures, one low
+  sub-bass boom.
+- Close with "no music, no melody, no song, no voice".
+- Put each transition in the sound design — a whoosh on every "cut to" — not in a
+  musical cue. A cue tends to restart at the cut, which is the discontinuity you
+  are avoiding.
+
+Dialogue is not affected by this: quoted lines are the model's own voice track,
+and the model-line skill says how to write them.
+
+## Check what came back
+
+- **Length.** `duration_seconds` is honoured loosely and clamped to the lengths a
+  model supports. Measure with `analyze_video` before cutting to it.
+- **Frame.** Aspect ratio is not guaranteed — an image-to-video route often
+  ignores a 9:16 source and emits 16:9. `analyze_video` reports it; fix it with
+  an ffmpeg crop to the target frame (`nodetool.media.ffmpeg` in a code action;
+  `api-media` has its contract) before compositing, not after.
+- **The mix.** Judge it with `understand_video` (Gemini gets the audio; every
+  other vision model is sent silent stills), `analyze_audio` for the levels, and
+  `detect_video_scenes` to confirm the cuts landed where the beats said. Never
+  from a contact sheet.
+- **The file.** Build on a stored asset. `generate_video` saves its result;
+  probing a generation node with `invoke_node` hands back whatever the node
+  returned, which can be a run-local path that is gone by the time you assemble.
+
+## Where the rest picks up
+
+`timeline-edit-ops` carries the timeline op contract for laying the bed and muting
+shot audio under a narration track. `beat-sync-editing` sits the cuts on that
+bed once it exists, and `caption-titles` times the type to it — a title card is
+never rendered into the clip, whichever build you chose.

package/dist/_skills/video-clone/SKILL.md

@@ -0,0 +1,96 @@
+---
+name: video-clone
+description: "Analyze a reference video and recreate its shot structure with a new product or cast in NodeTool."
+---
+
+**Load `/storyboard-core` first.** It carries the loop, the tool contract and
+the gating this skill assumes. Everything below is the breakdown pass that comes
+before that loop, and the rules for what may and may not be carried over.
+
+## What makes this job type its own
+
+The structure is given, so the work is reading it accurately and then rebuilding it
+with different content. Two failure modes: describing the reference vaguely enough
+that the rebuild is unrelated, and copying it closely enough that you reproduce
+somebody's trademarks.
+
+## Read the reference first
+
+`find_model` with `capability: "generate_message"` for a model that reads video —
+Gemini does it natively — then:
+
+```
+understand_video {provider, model, video, prompt, max_tokens}
+```
+
+`video` takes an asset id, an `asset://` URI, a URL or a data URI. It reads the whole
+video, not sampled frames. Ask for one thing at a time rather than everything at once;
+two or three calls read better than one overloaded prompt.
+
+Extract, in this order:
+
+1. **Frame:** aspect, total length, shot count.
+2. **The first three seconds** — what the hook actually does, described as an action.
+3. **Per shot:** in and out timecode, framing, camera height, camera move, who speaks,
+   whether the product is visible and how.
+4. **Audio recipe:** dialogue, voiceover, room tone, music, and whether the sound is
+   diegetic.
+5. **One paragraph on why it works** — the mechanism, not praise.
+
+Report the breakdown. Stop for an analysis-only request or a requested review.
+For an authorized remake, continue using the reference structure and the new brief.
+
+## What carries over, and what must not
+
+| Carry over | Replace |
+|---|---|
+| Cut rhythm — each shot's duration | The room, the set, the location |
+| Camera heights and framings | Wardrobe and casting |
+| The order of beats: hook, demonstration, verdict | Every spoken line |
+| Whether the sound is diegetic | The product |
+| Aspect ratio and total length | The palette, unless the user asks to match it |
+
+**Never carry a trademark across.** Logos, wordmarks, brand colours as identity,
+slogans, packaging layout and any readable sign in the reference stay out of the
+rebuild. Put `no brand marks other than [PRODUCT]` in the board `style`. If the user
+asks for the reference's actual logo, say plainly that the rebuild uses their product
+and nothing else, and carry on with the rest.
+
+## Rebuild
+
+Map each extracted shot to one `add_shot`, keeping its duration as
+`duration_seconds`, its framing in `camera.framing`, its camera height in
+`camera.angle` and its move in `camera.movement`. The still prompt reads framing,
+angle and lens, and the clip prompt reads the movement, so none of them needs
+repeating in `action`.
+
+Cast before rendering: a character entity from a generated reference you show the user
+first, and a prop entity from their pack shot. Slug the shots `1a`, `2a`, … so the
+user can name one against the reference's own numbering.
+
+Then run the core loop: stills, take selection (`select_version` when the user names a take), clips.
+
+## Brief
+
+Adapt these examples to the user's brief. Example deliverables, style choices,
+and approval checkpoints apply only when the user adopts them.
+
+```
+Read the attached reference video with understand_video (find a model that reads
+video via find_model, capability=generate_message — Gemini reads it natively).
+
+Extract: aspect, length, shot count; what the first three seconds do; then per
+shot the in/out timecode, framing, camera height, move, who speaks and whether
+the product is visible. Then the audio recipe, then one paragraph on the
+mechanism that makes it work.
+
+Report that and stop.
+
+Then rebuild it for [PRODUCT] with a new character. Keep the cut rhythm, the
+durations and the camera heights. Change the room, the wardrobe and every line.
+Carry no trademark from the reference — board style gets "no brand marks other
+than [PRODUCT]".
+
+Cast from stills first: generate the character reference, show me, then tag it.
+Slug every shot so I can say "fix 2a".
+```

package/dist/_skills/video-workflow/SKILL.md

@@ -0,0 +1,119 @@
+---
+name: video-workflow
+description: "Build a reusable NodeTool video workflow for batches or repeated inputs. For a single finished piece, use a storyboard skill."
+---
+
+**Load `/storyboard-core` for what a shot is and how entities season a prompt,
+and `/nodetool-workflow-builder` for how to author a graph at all** — node search,
+handles, validation, the tools. This skill is the seam between them: the nodes that
+turn a brief into shots, and when a graph is the wrong answer.
+
+## First, check a graph is actually wanted
+
+A graph earns its cost only when the pipeline gets **re-run on new inputs**. One video,
+even a complicated one, is faster and cheaper to direct on a storyboard — no node
+search, no handle wiring, no validation pass, and the user can see and fix each shot.
+
+Build a graph when the user says template, pipeline, batch, factory, "for every SKU",
+"run it again next month", or wants it callable from Chat by name. If they asked for a
+video and you are reaching for a graph, go back to the storyboard skills.
+
+## The creative nodes
+
+Search the registry before adding any of these — `search_nodes` or `ui_search_nodes`
+with `include_properties` and `include_outputs` — and never invent a type, a property
+or a handle name. Verified types:
+
+| Node | Does |
+|---|---|
+| `nodetool.creative.Director` | brief → `Screenplay`. Props: `model` (language_model), `brief`, `style`, `shot_count` (1–20), `aspect_ratio`, `max_tokens`. Outputs: `screenplay`, `narration`, `music_prompt`, `title`. |
+| `nodetool.creative.ScreenplayShots` | fans a screenplay out one shot at a time. Outputs `shot`, `shot_prompt`, `index` per iteration plus `output` as the whole list. |
+| `nodetool.creative.ShotBatch` | flattens a screenplay into generation-ready specs in one go. Output: `shots` (list[dict]) with prompt, timing and keyframe. |
+| `nodetool.creative.ShotChain` | animates a spec list sequentially, seeding each shot's first frame from the previous clip's last frame. Output: `videos`. Needs the `ffmpeg` runtime. |
+| `nodetool.creative.ApplyEntities` | seasons a prompt with entity descriptors, the same rule the storyboard render path uses. |
+| `nodetool.image.TextToImage` | prompt → image. |
+| `nodetool.video.ImageToVideo` / `nodetool.video.TextToVideo` | still → clip, or prompt → clip. |
+
+## Re-run an approved document instead of re-directing
+
+When the user already has a board, a script or a cut they signed off, do not put
+`Director` in the graph. Start from the approved row and derive:
+
+| Node | Does |
+|---|---|
+| `nodetool.constant.Storyboard` / `.Entity` | a picked board or library entity as a graph value. Outputs `output`. |
+| `nodetool.entity.CreateEntity` | `image`, `kind`, `name`, `descriptor`, `key` → `entity`, `created`. Upserts on `key` (a SKU), so a second run reuses the row instead of growing the library. |
+| `nodetool.entity.LoadEntity` / `ListEntities` | read one entity, or stream a filtered set. |
+| `nodetool.storyboard.CreateStoryboard` | `screenplay` from `Director`, `name`, `cast`, both models → a new writable `storyboard`. The brief-first path: Director → CreateStoryboard → RenderStills. A re-run returns the board it made. |
+| `nodetool.storyboard.RecastStoryboard` | `storyboard`, `cast`, `replaces`, `reuse_existing` → the copy, `invalidated`, `kept`. Keeps every frame whose prompt did not move. |
+| `nodetool.storyboard.RenderStills` / `RenderClips` | render the stale shots of a derived board. `only_stale`, `max_shots`, `require_keyframe` are the spend gates. |
+| `nodetool.storyboard.AssembleTimeline` | the derived board → `timeline`. A copy inherits the template's cut, titles and music. |
+| `nodetool.storyboard.LoadStoryboard` / `StoryboardShots` | read a board's shots, entities and settings, or stream one message per shot with its still and clip. |
+| `nodetool.script.WriteScript` / `FillScript` | write a script from a brief and a cast, or fill `{{key}}` placeholders in an approved one. Both make a new row. |
+| `nodetool.timeline.FillTimelineText` | `timeline`, `values` → a new sequence with the overlay filled. |
+| `nodetool.timeline.RetargetTimeline` | `timeline`, `aspect_ratio`, `fit` → a new sequence, plus `cropped` naming the clips the crop cuts into. |
+
+**The write contract.** A `storyboard` ref is read-only unless the run derived it.
+`RenderStills`, `RenderClips` and `AssembleTimeline` refuse a picked board and name
+`allow_writes`, which is the override for a graph whose whole purpose is to render the
+board a person chose. Wire `CreateStoryboard` or `RecastStoryboard` first and pass its
+output along; that ref carries the permission.
+
+Shipped examples to read before wiring one yourself: **Per-SKU Ad Factory**,
+**Localized Explainer**, **Three Ratios** (`get_example_workflow`).
+
+## Two shapes, and they are not interchangeable
+
+**Per-shot fan-out** — `Director` → `ScreenplayShots` → `TextToImage` → `ImageToVideo`.
+Each shot renders independently. Use it when shots are separate takes: an ad, a set of
+social cuts, anything where a shot failing should not spoil the rest.
+
+**Continuous chain** — `Director` → `ShotBatch` → `ShotChain`. Each clip's first frame
+is the previous clip's last frame, so motion and framing carry across the cut. Use it
+for a sequence that has to feel continuous. It is sequential, so it is slower and one
+bad clip propagates.
+
+`ApplyEntities` goes between the prompt source and the image node whenever the pipeline
+has a recurring character or product to hold steady.
+
+## Wire the gate in
+
+The storyboard loop's value is that a human sees the stills before the clips are paid
+for. A graph loses that unless you build it in: keep the still half and the clip half
+separable, so the user can run to the stills, look, and then run the clip half. Say
+which inputs do that when you hand the workflow over.
+
+## Validate and deliver the requested result
+
+`validate_workflow` with the inline graph before saving — it catches unknown node
+types, missing required properties, unselected models, dangling edges and a model id
+the provider does not offer, in well under a second and with no spend.
+
+`create_workflow` refuses to save a graph whose model properties are unselected, so
+pick real ones with `find_model` and stamp them in. Nothing fills them in at run time.
+
+Save and read back the graph. A request to build a template ends with a validated
+workflow and its input instructions. If the user also requested execution, run
+the authorized stages within their scope and budget. Do not request the same
+authorization again.
+
+## Brief
+
+Adapt these examples to the user's brief. Example deliverables, style choices,
+and approval checkpoints apply only when the user adopts them.
+
+```
+Build a workflow named "UGC factory" that I can re-run from Chat.
+
+Inputs: a product image, a brief, an aspect ratio.
+Shape: nodetool.creative.Director (brief + style + shot_count) ->
+nodetool.creative.ScreenplayShots -> nodetool.creative.ApplyEntities ->
+nodetool.image.TextToImage -> a point I can stop at and look ->
+nodetool.video.ImageToVideo -> timeline assembly.
+
+Search the registry for every node type and read its real properties and handles
+before you add it. Pick concrete models with find_model and stamp them in.
+validate_workflow before saving.
+
+Save it and stop. Do not run the clip stage.
+```

package/dist/_skills/wan-2-6-prompting/SKILL.md

@@ -0,0 +1,107 @@
+---
+name: wan-2-6-prompting
+description: Prompt Alibaba's Wan 2.6 video — text-to-video with a global style line plus timing-bracketed shots, and image-to-video that describes only what changes over time. Covers the shot-mode switch per provider, the prompt budget, negative prompts, motion intensity wording, audio input and prompt expansion. Use whenever the model id names a Wan 2.6 video route (alibaba/wan-2.6/text-to-video, alibaba/wan-2.6/image-to-video, wan-video/wan-2.6-t2v, wan-video/wan-2.6-i2v, wan-video/wan2.6-i2v-flash) on generate_video, animate_image or a TextToVideo node. Not for Wan 2.5 or 2.7, or the Wan 2.6 image routes.
+---
+
+# Wan 2.6 → two modes, two prompt shapes
+
+Wan 2.6 generates from text or animates a still. Each mode wants a differently
+shaped prompt. Getting the shape wrong is the usual reason a clip comes back
+generic.
+
+Reach it with `find_model` for `text_to_video` or `image_to_video`, then
+`generate_video` / `animate_image`. Those forward the prompt (and the image),
+a negative prompt (not `animate_image`), the duration, the aspect ratio and
+the resolution. The shot-mode switch, prompt expansion and an audio file exist
+only on the route's own node: find it with `search_nodes`, read its fields
+with `get_node_info`, and run it with `run_node`. Each run is a paid
+generation, so confirm the scope with the user before rendering several.
+
+fal's guide caps main prompts at 800 characters. No shipped route states the
+limit, but write as if it holds.
+
+## Text to video
+
+Two components: a global style line, then shots with timing brackets.
+
+```
+A cinematic journey through ancient ruins at sunset. Photoreal, 4K, film grain.
+
+Shot 1 [0-3s] Wide establishing shot of stone pillars with sunlight streaming through.
+Shot 2 [3-7s] Camera tracks forward through an archway revealing a hidden chamber.
+Shot 3 [7-10s] Close-up of ancient inscriptions as dust particles float in light beams.
+```
+
+The shot mode is a switch, and its name and default differ by provider:
+AtlasCloud's `shot_type` defaults to `multi`, and Replicate's `multi_shots`
+defaults to false, so set it true there before sending bracketed shots. Give
+each shot a timing indicator, a
+camera action (push, pull, pan, orbit, track) and its scene elements — subject
+position, lighting change, environmental detail. Keep continuity by referencing
+the same characters, locations and objects across shots; unrelated shots
+produce disjointed results.
+
+AtlasCloud's durations are 5, 10 or 15 seconds, at 720p or 1080p, in 16:9, 9:16, 1:1, 4:3 or
+3:4. Write to the ratio you picked: wide establishing shots and horizontal
+movement for 16:9, tighter framing and vertical composition for 9:16, centred
+subjects for 1:1.
+
+## Image to video
+
+Describe the temporal change, not the image. The model already has the frame;
+words spent re-describing it are words not spent on motion.
+
+```
+Continue from first frame. Gentle camera push toward the mountain peak as
+clouds drift overhead. Light changes from morning to golden hour. Cinematic and
+serene movement.
+```
+
+Spend the prompt on camera motion, lighting shifts, and environmental
+animation — water, clouds, foliage, smoke. The same shot switch applies:
+AtlasCloud defaults to `multi`, Replicate to `multi_shots` false. Set single
+shot for a plain animated still.
+
+Images animate better when they are high resolution, have clear depth of field,
+carry atmospheric elements, and are uncluttered with a well-defined subject.
+Busy compositions with competing elements produce inconsistent motion. fal's
+guide gives input images 360 to 2000 px per side, up to 100 MB.
+
+## Controls
+
+**Negative prompt** — fal's guide caps it at 500 characters. Prioritise the artifacts you actually
+see: "low quality, blurry, distorted faces, unnatural movement, text,
+watermarks, shaky camera".
+
+**Motion intensity** is a wording choice, not a parameter. Minimal: "subtle
+camera drift", "gentle movement". Moderate: "smooth camera track", "flowing
+motion". Dramatic: "dynamic sweeping motion", "rapid camera movement".
+
+**Audio input** (`audio` on the route's node) takes WAV or MP3, 3 to 30
+seconds, up to 15 MB, per fal's guide. Audio longer than the video is
+truncated; audio shorter leaves the rest silent. For
+dialogue, put speaker cues in the prompt.
+
+**Prompt expansion** (`enable_prompt_expansion`) is on by default on every
+Wan 2.6 video route and runs an LLM over the prompt before generation. It adds
+detail without eating the 800-character budget, and it
+works best when you have given it style references, visual descriptors and
+precise terminology to expand from.
+
+## Symptoms
+
+| What went wrong | What to change |
+| :--- | :--- |
+| Incoherent narrative | Break the scene into specific shots with timing brackets |
+| Inconsistent subjects | Repeat the character description across every shot |
+| Unnatural motion | Name the camera movement explicitly: push, pan, orbit |
+| Flat visual quality | Add quality descriptors to the global style line |
+| An element ignored | Move it earlier in the shot description |
+
+Iterate one variable at a time, and start in text-to-video — the shot and
+timing format transfers to image-to-video without any asset dependency.
+Check the render with `analyze_video`, `detect_video_scenes` and
+`understand_video`.
+
+Adapted from fal's Wan 2.6 prompt guide:
+https://fal.ai/learn/devs/wan-2-6-prompt-guide-mastering-all-three-generation-modes

package/dist/analysis/audio-dsp.d.ts

@@ -12,6 +12,7 @@
  * streaming target is specified in. dBFS peak/RMS is kept alongside it because
  * clipping and headroom are peak questions, not loudness questions.
  */
+export { detectOnsets, estimateTempo, fftInPlace, hannWindow, magnitudeSpectrum, spectralFlux, type TempoEstimate } from "@nodetool-ai/timeline";
 /** Amplitude floor reported instead of -Infinity for digital silence. */
 export declare const SILENT_DB = -120;
 /** Interleaved PCM plus the format needed to interpret it. */
@@ -63,20 +64,6 @@ export interface PeakSummary {
 export declare function peakSummary(mono: Float32Array, clipThreshold?: number): PeakSummary;
 /** Largest power of two at or below `value`, at least 2. */
 export declare function floorPowerOfTwo(value: number): number;
-/** A periodic Hann window of length `size`. */
-export declare function hannWindow(size: number): Float32Array;
-/**
- * In-place radix-2 Cooley-Tukey FFT. `re`/`im` must have the same
- * power-of-two length; the transform overwrites both.
- */
-export declare function fftInPlace(re: Float32Array, im: Float32Array): void;
-/**
- * The magnitude spectrum of one windowed frame — bins 0..size/2 inclusive.
- *
- * `signal` is read from `offset`; a frame that runs past the end is
- * zero-padded, which is what makes the last window of a clip analysable.
- */
-export declare function magnitudeSpectrum(signal: Float32Array, offset: number, window: Float32Array): Float32Array;
 /** What one spectrum says about the sound's colour. */
 export interface SpectralFeatures {
     /** Energy-weighted mean frequency — "brightness". */
@@ -158,21 +145,6 @@ export interface LoudnessResult {
  * number nobody measured.
  */
 export declare function measureLoudness(planar: readonly Float32Array[], sampleRate: number): LoudnessResult;
-/**
- * Half-wave rectified spectral flux between two frames, normalized by the
- * current frame's total magnitude.
- *
- * The normalization is what makes the number mean something. Raw flux scales
- * with how loud the passage is, so one threshold cannot serve a quiet verse
- * and a loud chorus. Worse, a *stationary* tone does not produce zero raw
- * flux: at a hop that is not a whole number of cycles the sinusoid's leakage
- * pattern shifts between frames, so its bin magnitudes wobble by a few percent
- * forever. Measured against the frame's own magnitude that wobble is ~0.02 and
- * a real onset is a large fraction of 1, which is a difference a threshold can
- * act on. Against raw flux the two are indistinguishable, and a held note
- * reads as a stream of onsets at a confident tempo.
- */
-export declare function spectralFlux(magnitude: Float32Array, previous: Float32Array): number;
 /** A stretch of the timeline classified as silent or as sound. */
 export interface AudioSegment {
     readonly start: number;
@@ -190,35 +162,4 @@ export interface AudioSegment {
 export declare function silenceSegments(frames: readonly EnergyFrame[], thresholdDb: number, minDurationSeconds: number, frameDuration: number): AudioSegment[];
 /** The gaps between silences: everything that is not silent. */
 export declare function invertSegments(segments: readonly AudioSegment[], duration: number): AudioSegment[];
-/**
- * Onset times from a normalized spectral-flux novelty curve.
- *
- * A peak is an onset when it is a local maximum, clears an adaptive threshold
- * (the local mean plus `sensitivity` local standard deviations), and clears
- * `minStrength` outright. Both halves are needed. The adaptive half handles
- * dynamics — a fixed threshold finds every hit in a loud passage and none in a
- * quiet one. The absolute half handles the opposite failure: where the curve
- * is essentially flat the local standard deviation collapses, every ripple
- * clears mean + 1.5σ, and a held note reports dozens of onsets. `minStrength`
- * is in the units {@link spectralFlux} returns — a fraction of the frame's own
- * magnitude — so 0.04 means "at least 4% of this frame's energy is new".
- */
-export declare function detectOnsets(flux: readonly number[], times: readonly number[], sensitivity?: number, windowFrames?: number, minGapSeconds?: number, minStrength?: number): number[];
-/** A tempo estimate, and how strongly the novelty curve supports it. */
-export interface TempoEstimate {
-    readonly bpm: number;
-    /** Normalized autocorrelation at the winning lag, 0..1. */
-    readonly confidence: number;
-}
-/**
- * Tempo from the autocorrelation of the novelty curve.
- *
- * Only lags inside `minBpm`..`maxBpm` are considered, so the classic
- * half/double-time ambiguity is bounded rather than unbounded. Confidence is
- * the autocorrelation at the winning lag over the curve's own energy; a
- * non-rhythmic signal scores near zero, and the caller is expected to say so
- * instead of printing a BPM for speech.
- */
-export declare function estimateTempo(novelty: readonly number[], hopSeconds: number, minBpm?: number, maxBpm?: number): TempoEstimate;
-export {};
 //# sourceMappingURL=audio-dsp.d.ts.map

package/dist/analysis/audio-dsp.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"audio-dsp.d.ts","sourceRoot":"","sources":["../../src/analysis/audio-dsp.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AAEH,yEAAyE;AACzE,eAAO,MAAM,SAAS,OAAO,CAAC;AAE9B,8DAA8D;AAC9D,MAAM,WAAW,QAAQ;IACvB,QAAQ,CAAC,OAAO,EAAE,YAAY,CAAC;IAC/B,QAAQ,CAAC,UAAU,EAAE,MAAM,CAAC;IAC5B,QAAQ,CAAC,QAAQ,EAAE,MAAM,CAAC;CAC3B;AAED,mEAAmE;AACnE,wBAAgB,aAAa,CAAC,SAAS,EAAE,MAAM,GAAG,MAAM,CAGvD;AAED,kEAAkE;AAClE,wBAAgB,KAAK,CAAC,KAAK,EAAE,MAAM,EAAE,MAAM,SAAI,GAAG,MAAM,CAIvD;AAED,2DAA2D;AAC3D,wBAAgB,MAAM,CAAC,OAAO,EAAE,YAAY,EAAE,QAAQ,EAAE,MAAM,GAAG,YAAY,CAY5E;AAED,wDAAwD;AACxD,wBAAgB,QAAQ,CACtB,OAAO,EAAE,YAAY,EACrB,QAAQ,EAAE,MAAM,GACf,YAAY,EAAE,CAWhB;AAMD,gFAAgF;AAChF,MAAM,WAAW,WAAW;IAC1B,gEAAgE;IAChE,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,QAAQ,CAAC,GAAG,EAAE,MAAM,CAAC;IACrB,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;CACvB;AAED;;;;;;;;GAQG;AACH,wBAAgB,YAAY,CAC1B,OAAO,EAAE,YAAY,EACrB,UAAU,EAAE,MAAM,EAClB,YAAY,EAAE,MAAM,EACpB,UAAU,EAAE,MAAM,EAClB,QAAQ,SAAI,GACX,WAAW,EAAE,CAyBf;AAED,2EAA2E;AAC3E,MAAM,WAAW,WAAW;IAC1B,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,QAAQ,CAAC,GAAG,EAAE,MAAM,CAAC;IACrB,QAAQ,CAAC,QAAQ,EAAE,MAAM,CAAC;IAC1B,QAAQ,CAAC,cAAc,EAAE,MAAM,CAAC;IAChC,QAAQ,CAAC,aAAa,EAAE,MAAM,CAAC;CAChC;AAED;;;;;;GAMG;AACH,wBAAgB,WAAW,CACzB,IAAI,EAAE,YAAY,EAClB,aAAa,SAAQ,GACpB,WAAW,CAsBb;AAMD,4DAA4D;AAC5D,wBAAgB,eAAe,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM,CAIrD;AAED,+CAA+C;AAC/C,wBAAgB,UAAU,CAAC,IAAI,EAAE,MAAM,GAAG,YAAY,CAMrD;AAED;;;GAGG;AACH,wBAAgB,UAAU,CAAC,EAAE,EAAE,YAAY,EAAE,EAAE,EAAE,YAAY,GAAG,IAAI,CA2CnE;AAED;;;;;GAKG;AACH,wBAAgB,iBAAiB,CAC/B,MAAM,EAAE,YAAY,EACpB,MAAM,EAAE,MAAM,EACd,MAAM,EAAE,YAAY,GACnB,YAAY,CAcd;AAED,uDAAuD;AACvD,MAAM,WAAW,gBAAgB;IAC/B,qDAAqD;IACrD,QAAQ,CAAC,UAAU,EAAE,MAAM,CAAC;IAC5B,kEAAkE;IAClE,QAAQ,CAAC,SAAS,EAAE,MAAM,CAAC;IAC3B,mEAAmE;IACnE,QAAQ,CAAC,QAAQ,EAAE,MAAM,CAAC;IAC1B,kDAAkD;IAClD,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;IAC7B,mCAAmC;IACnC,QAAQ,CAAC,MAAM,EAAE,MAAM,CAAC;IACxB,4DAA4D;IAC5D,QAAQ,CAAC,MAAM,EAAE,MAAM,CAAC;CACzB;AAED,2EAA2E;AAC3E,wBAAgB,gBAAgB,CAC9B,SAAS,EAAE,YAAY,EACvB,UAAU,EAAE,MAAM,EAClB,OAAO,EAAE,MAAM,EACf,eAAe,SAAO,GACrB,gBAAgB,CAuDlB;AAED,0DAA0D;AAC1D,MAAM,WAAW,aAAa;IAC5B,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,QAAQ,CAAC,KAAK,EAAE,MAAM,CAAC;IACvB,QAAQ,CAAC,MAAM,EAAE,MAAM,CAAC;CACzB;AAED;;;GAGG;AACH,eAAO,MAAM,YAAY,EAAE,SAAS,aAAa,EAWhD,CAAC;AAEF,4EAA4E;AAC5E,wBAAgB,YAAY,CAC1B,SAAS,EAAE,YAAY,EACvB,UAAU,EAAE,MAAM,EAClB,OAAO,EAAE,MAAM,GACd;IAAE,IAAI,EAAE,aAAa,CAAC;IAAC,MAAM,EAAE,MAAM,CAAA;CAAE,EAAE,CAe3C;AAMD,UAAU,kBAAkB;IAC1B,QAAQ,CAAC,EAAE,EAAE,MAAM,CAAC;IACpB,QAAQ,CAAC,EAAE,EAAE,MAAM,CAAC;IACpB,QAAQ,CAAC,EAAE,EAAE,MAAM,CAAC;IACpB,QAAQ,CAAC,EAAE,EAAE,MAAM,CAAC;IACpB,QAAQ,CAAC,EAAE,EAAE,MAAM,CAAC;CACrB;AAED,8EAA8E;AAC9E,wBAAgB,WAAW,CACzB,MAAM,EAAE,YAAY,EACpB,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,kBAAkB,GACzC,YAAY,CAgBd;AAED;;;;;;;;GAQG;AACH,wBAAgB,iBAAiB,CAC/B,UAAU,EAAE,MAAM,GACjB,CAAC,kBAAkB,EAAE,kBAAkB,CAAC,CA4B1C;AAWD,4CAA4C;AAC5C,MAAM,WAAW,cAAc;IAC7B,+DAA+D;IAC/D,QAAQ,CAAC,cAAc,EAAE,MAAM,CAAC;IAChC,2CAA2C;IAC3C,QAAQ,CAAC,eAAe,EAAE,MAAM,CAAC;IACjC,qCAAqC;IACrC,QAAQ,CAAC,gBAAgB,EAAE,MAAM,CAAC;IAClC,kCAAkC;IAClC,QAAQ,CAAC,gBAAgB,EAAE,MAAM,CAAC;IAClC,mEAAmE;IACnE,QAAQ,CAAC,SAAS,EAAE;QAAE,IAAI,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAA;KAAE,EAAE,CAAC;CACtD;AAkCD;;;;;;;;;;;GAWG;AACH,wBAAgB,eAAe,CAC7B,MAAM,EAAE,SAAS,YAAY,EAAE,EAC/B,UAAU,EAAE,MAAM,GACjB,cAAc,CAwGhB;AAMD;;;;;;;;;;;;;GAaG;AACH,wBAAgB,YAAY,CAC1B,SAAS,EAAE,YAAY,EACvB,QAAQ,EAAE,YAAY,GACrB,MAAM,CASR;AAED,kEAAkE;AAClE,MAAM,WAAW,YAAY;IAC3B,QAAQ,CAAC,KAAK,EAAE,MAAM,CAAC;IACvB,QAAQ,CAAC,GAAG,EAAE,MAAM,CAAC;IACrB,QAAQ,CAAC,QAAQ,EAAE,MAAM,CAAC;CAC3B;AAED;;;;;;;GAOG;AACH,wBAAgB,eAAe,CAC7B,MAAM,EAAE,SAAS,WAAW,EAAE,EAC9B,WAAW,EAAE,MAAM,EACnB,kBAAkB,EAAE,MAAM,EAC1B,aAAa,EAAE,MAAM,GACpB,YAAY,EAAE,CAsBhB;AAED,gEAAgE;AAChE,wBAAgB,cAAc,CAC5B,QAAQ,EAAE,SAAS,YAAY,EAAE,EACjC,QAAQ,EAAE,MAAM,GACf,YAAY,EAAE,CAYhB;AAED;;;;;;;;;;;;GAYG;AACH,wBAAgB,YAAY,CAC1B,IAAI,EAAE,SAAS,MAAM,EAAE,EACvB,KAAK,EAAE,SAAS,MAAM,EAAE,EACxB,WAAW,SAAM,EACjB,YAAY,SAAK,EACjB,aAAa,SAAO,EACpB,WAAW,SAAO,GACjB,MAAM,EAAE,CA4BV;AAED,wEAAwE;AACxE,MAAM,WAAW,aAAa;IAC5B,QAAQ,CAAC,GAAG,EAAE,MAAM,CAAC;IACrB,2DAA2D;IAC3D,QAAQ,CAAC,UAAU,EAAE,MAAM,CAAC;CAC7B;AAED;;;;;;;;GAQG;AACH,wBAAgB,aAAa,CAC3B,OAAO,EAAE,SAAS,MAAM,EAAE,EAC1B,UAAU,EAAE,MAAM,EAClB,MAAM,SAAK,EACX,MAAM,SAAM,GACX,aAAa,CA8Bf"}
+{"version":3,"file":"audio-dsp.d.ts","sourceRoot":"","sources":["../../src/analysis/audio-dsp.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AAIH,OAAO,EACL,YAAY,EACZ,aAAa,EACb,UAAU,EACV,UAAU,EACV,iBAAiB,EACjB,YAAY,EACZ,KAAK,aAAa,EACnB,MAAM,uBAAuB,CAAC;AAE/B,yEAAyE;AACzE,eAAO,MAAM,SAAS,OAAO,CAAC;AAE9B,8DAA8D;AAC9D,MAAM,WAAW,QAAQ;IACvB,QAAQ,CAAC,OAAO,EAAE,YAAY,CAAC;IAC/B,QAAQ,CAAC,UAAU,EAAE,MAAM,CAAC;IAC5B,QAAQ,CAAC,QAAQ,EAAE,MAAM,CAAC;CAC3B;AAED,mEAAmE;AACnE,wBAAgB,aAAa,CAAC,SAAS,EAAE,MAAM,GAAG,MAAM,CAGvD;AAED,kEAAkE;AAClE,wBAAgB,KAAK,CAAC,KAAK,EAAE,MAAM,EAAE,MAAM,SAAI,GAAG,MAAM,CAIvD;AAED,2DAA2D;AAC3D,wBAAgB,MAAM,CAAC,OAAO,EAAE,YAAY,EAAE,QAAQ,EAAE,MAAM,GAAG,YAAY,CAY5E;AAED,wDAAwD;AACxD,wBAAgB,QAAQ,CACtB,OAAO,EAAE,YAAY,EACrB,QAAQ,EAAE,MAAM,GACf,YAAY,EAAE,CAWhB;AAMD,gFAAgF;AAChF,MAAM,WAAW,WAAW;IAC1B,gEAAgE;IAChE,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,QAAQ,CAAC,GAAG,EAAE,MAAM,CAAC;IACrB,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;CACvB;AAED;;;;;;;;GAQG;AACH,wBAAgB,YAAY,CAC1B,OAAO,EAAE,YAAY,EACrB,UAAU,EAAE,MAAM,EAClB,YAAY,EAAE,MAAM,EACpB,UAAU,EAAE,MAAM,EAClB,QAAQ,SAAI,GACX,WAAW,EAAE,CAyBf;AAED,2EAA2E;AAC3E,MAAM,WAAW,WAAW;IAC1B,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,QAAQ,CAAC,GAAG,EAAE,MAAM,CAAC;IACrB,QAAQ,CAAC,QAAQ,EAAE,MAAM,CAAC;IAC1B,QAAQ,CAAC,cAAc,EAAE,MAAM,CAAC;IAChC,QAAQ,CAAC,aAAa,EAAE,MAAM,CAAC;CAChC;AAED;;;;;;GAMG;AACH,wBAAgB,WAAW,CACzB,IAAI,EAAE,YAAY,EAClB,aAAa,SAAQ,GACpB,WAAW,CAsBb;AAMD,4DAA4D;AAC5D,wBAAgB,eAAe,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM,CAIrD;AAED,uDAAuD;AACvD,MAAM,WAAW,gBAAgB;IAC/B,qDAAqD;IACrD,QAAQ,CAAC,UAAU,EAAE,MAAM,CAAC;IAC5B,kEAAkE;IAClE,QAAQ,CAAC,SAAS,EAAE,MAAM,CAAC;IAC3B,mEAAmE;IACnE,QAAQ,CAAC,QAAQ,EAAE,MAAM,CAAC;IAC1B,kDAAkD;IAClD,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;IAC7B,mCAAmC;IACnC,QAAQ,CAAC,MAAM,EAAE,MAAM,CAAC;IACxB,4DAA4D;IAC5D,QAAQ,CAAC,MAAM,EAAE,MAAM,CAAC;CACzB;AAED,2EAA2E;AAC3E,wBAAgB,gBAAgB,CAC9B,SAAS,EAAE,YAAY,EACvB,UAAU,EAAE,MAAM,EAClB,OAAO,EAAE,MAAM,EACf,eAAe,SAAO,GACrB,gBAAgB,CAuDlB;AAED,0DAA0D;AAC1D,MAAM,WAAW,aAAa;IAC5B,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,QAAQ,CAAC,KAAK,EAAE,MAAM,CAAC;IACvB,QAAQ,CAAC,MAAM,EAAE,MAAM,CAAC;CACzB;AAED;;;GAGG;AACH,eAAO,MAAM,YAAY,EAAE,SAAS,aAAa,EAWhD,CAAC;AAEF,4EAA4E;AAC5E,wBAAgB,YAAY,CAC1B,SAAS,EAAE,YAAY,EACvB,UAAU,EAAE,MAAM,EAClB,OAAO,EAAE,MAAM,GACd;IAAE,IAAI,EAAE,aAAa,CAAC;IAAC,MAAM,EAAE,MAAM,CAAA;CAAE,EAAE,CAe3C;AAMD,UAAU,kBAAkB;IAC1B,QAAQ,CAAC,EAAE,EAAE,MAAM,CAAC;IACpB,QAAQ,CAAC,EAAE,EAAE,MAAM,CAAC;IACpB,QAAQ,CAAC,EAAE,EAAE,MAAM,CAAC;IACpB,QAAQ,CAAC,EAAE,EAAE,MAAM,CAAC;IACpB,QAAQ,CAAC,EAAE,EAAE,MAAM,CAAC;CACrB;AAED,8EAA8E;AAC9E,wBAAgB,WAAW,CACzB,MAAM,EAAE,YAAY,EACpB,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,kBAAkB,GACzC,YAAY,CAgBd;AAED;;;;;;;;GAQG;AACH,wBAAgB,iBAAiB,CAC/B,UAAU,EAAE,MAAM,GACjB,CAAC,kBAAkB,EAAE,kBAAkB,CAAC,CA4B1C;AAWD,4CAA4C;AAC5C,MAAM,WAAW,cAAc;IAC7B,+DAA+D;IAC/D,QAAQ,CAAC,cAAc,EAAE,MAAM,CAAC;IAChC,2CAA2C;IAC3C,QAAQ,CAAC,eAAe,EAAE,MAAM,CAAC;IACjC,qCAAqC;IACrC,QAAQ,CAAC,gBAAgB,EAAE,MAAM,CAAC;IAClC,kCAAkC;IAClC,QAAQ,CAAC,gBAAgB,EAAE,MAAM,CAAC;IAClC,mEAAmE;IACnE,QAAQ,CAAC,SAAS,EAAE;QAAE,IAAI,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAA;KAAE,EAAE,CAAC;CACtD;AAkCD;;;;;;;;;;;GAWG;AACH,wBAAgB,eAAe,CAC7B,MAAM,EAAE,SAAS,YAAY,EAAE,EAC/B,UAAU,EAAE,MAAM,GACjB,cAAc,CAwGhB;AAMD,kEAAkE;AAClE,MAAM,WAAW,YAAY;IAC3B,QAAQ,CAAC,KAAK,EAAE,MAAM,CAAC;IACvB,QAAQ,CAAC,GAAG,EAAE,MAAM,CAAC;IACrB,QAAQ,CAAC,QAAQ,EAAE,MAAM,CAAC;CAC3B;AAED;;;;;;;GAOG;AACH,wBAAgB,eAAe,CAC7B,MAAM,EAAE,SAAS,WAAW,EAAE,EAC9B,WAAW,EAAE,MAAM,EACnB,kBAAkB,EAAE,MAAM,EAC1B,aAAa,EAAE,MAAM,GACpB,YAAY,EAAE,CAsBhB;AAED,gEAAgE;AAChE,wBAAgB,cAAc,CAC5B,QAAQ,EAAE,SAAS,YAAY,EAAE,EACjC,QAAQ,EAAE,MAAM,GACf,YAAY,EAAE,CAYhB"}