HEARD Platform Intelligence

How HEARD analyzes your music.

HEARD combines audio upload verification, measurement analysis where available, audio recognition, lyric transcription, and editorial intelligence to produce structured music reviews and artist content.

This page explains the analysis pipeline in consumer language. Internal service configuration is not exposed here.

Who uses HEARD's analysis pipeline

Independent Artists

Upload tracks and receive evidence-backed editorial reviews, artist interviews, and public profile pages grounded in real audio analysis.

Labels & A&R Teams

Run structured reviews across multiple artists, surface consistent measurements, and build catalog documentation for developing acts.

Press & Editorial Teams

Generate SEO-ready editorial articles, pull quotes, and press assets from structured review data — with source provenance for each claim.

The Analysis Pipeline

Five layers. Each one distinct.

Audio Upload & Verification

When you submit audio, HEARD verifies the file format, duration, and encoding before analysis begins. Supported formats include MP3, WAV, FLAC, and AIFF.

File integrity checked on upload
Duration and sample rate recorded
Source metadata extracted from file headers

DSP Measurements

When an external DSP service is connected and analysis is available, HEARD can surface measured values from your audio: tempo, key, loudness, dynamic range, and more.

Measured BPM and beat confidence
Key and scale detection
Integrated loudness (LUFS) where available
True peak (dBTP) where available
Spectral and energy characteristics
DSP measurements require a connected external audio analysis service. When unavailable, measurements are omitted — not estimated.

Audio Recognition

HEARD can attempt audio fingerprint recognition to match catalog metadata: artist name, title, ISRC, release date, label, and streaming links.

Fingerprint-based recognition (not AI invention)
Returns ISRC, UPC, and catalog metadata where matched
Surfaces Apple Music, Spotify, and MusicBrainz links when confirmed
Recognition matches must be confirmed — HEARD does not assert unverified catalog metadata as fact.

Lyric Transcription

When no lyrics are provided, HEARD can attempt audio transcription from speech. The result is shown to you for review and confirmation before being used in lyric analysis.

Transcription is internal — never published publicly
You review and confirm the transcript before it is used
Lyric analysis only runs on confirmed lyrics
You can skip transcription entirely and proceed as audio-only
Transcription accuracy varies by vocal clarity, mix density, and language. HEARD flags uncertainty rather than asserting wrong lyrics.

Editorial Intelligence

HEARD uses a multi-modal AI pipeline to produce editorial analysis from audio, lyrics, metadata, and artist-provided context. This is the core of the review writing layer.

Scores production, vocals, lyrics, structure, and replay value
Generates editorial body text, pull quotes, and press copy
Cross-checks claims against evidence before generating output
Originality scan runs before review is finalized
This layer produces interpreted editorial analysis, not objective measurements. Scores reflect editorial judgment — not absolute truth.

Two Distinct Systems

Track Library and Review Archive are not the same thing.

Artists sometimes confuse the two. Here is the difference: Track Library is the evidence layer. Review Archive is the output layer.

Review Archive

A list of your completed editorial reviews — the output layer. Each entry shows the final review, scores, press assets, and publication status. This is where you manage, edit, and share what HEARD produced.

Access: /archive (requires account)

Track Library

The persistent catalog of your uploaded audio tracks — the evidence layer. This is where DSP measurements, recognition matches, transcripts, analysis runs, and audio evidence attach to individual track records.

Access: /track-library (requires account)

What connects to a Track Library record

Track records

Each uploaded track gets a persistent record in your Track Library, including file metadata, duration, channels, and source details.

Audio evidence

DSP measurements, recognition results, and analysis-run data attach to the track record — making findings reusable across reviews, interviews, and profiles.

Browser analysis

When you play a track in the HEARD Live review mode, browser-observed evidence (waveform, energy shape, section markers) can also attach to the track.

Review connections

Completed reviews link back to their source track, allowing the platform to show analysis history and evidence provenance.

Interview connections

Interview context can reference track records — enabling interviewers to ask questions grounded in what HEARD actually measured about the song.

Profile connections

Catalog tracks can be surfaced on your public artist profile page, creating a documented body of work over time.

Share Studio

From review output to shareable assets.

Share Studio is the export layer inside HEARD. Once a review or interview is approved and published, Share Studio converts approved material into social-ready assets — score cards, pull quote cards, story formats, feed formats, and captions.

Score Card

1:1 Square

Clean export for feed posts

Story Card

9:16 Portrait

Optimized for Stories and Reels

Pull Quote

4:5 Feed

Single quote highlight format

Landscape

16:9 Wide

Header and banner format

Share Studio exports are based on approved published material only. Watermark-free exports require a paid subscription.

Artist Profiles

A public-facing portfolio built from your HEARD content.

Artist profiles are included for all HEARD users. The profile connects your approved reviews, published interviews, catalog tracks, photos, streaming links, and press copy into a single artist-facing public page — visible only when you choose to publish it.

Connects approved reviews and published interviews
Displays catalog tracks with audio evidence where available
Streaming and social links (Spotify, Apple Music, Instagram, and more)
Press copy and pull quotes from approved HEARD content
Profile photo, bio, and genre tags
Artist-controlled visibility — nothing public without your approval
Customization options vary by subscription plan
Public URL at /p/[your-slug]

What we do not guarantee

Honest boundaries about what the analysis pipeline can and cannot deliver.

DSP measurements require a connected external audio analysis service. They are not always available.
Audio recognition only returns confirmed fingerprint matches — unrecognized tracks remain unlabeled.
Lyric transcription accuracy varies by recording quality and is never used without your confirmation.
HEARD does not guarantee any specific measurement will be available for any given track.
Browser-observed evidence (waveform, energy) is exploratory — not equivalent to professional DSP measurement.
No external API keys, DSP service names, or provider credentials are exposed on public pages.

Ready to submit your first track?

2 complimentary review credits and 1 Mini Interview included on the free plan. No credit card required.