audio-calibration-mcp
Provides an optional separate bridge for Denon AVRs (via A1 Evo), enabling Denon integration without assuming specific AVR capabilities.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@audio-calibration-mcp Run audio doctor and plan a calibration session for my speakers"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Audio Calibration MCP
Audio Calibration MCP is a local-first, cross-platform Model Context Protocol server for Room EQ Wizard (REW). It guides acoustic measurement, speaker and room calibration, subwoofer and crossover analysis, conservative EQ, DSP deployment, and measured verification for general, powered-speaker, car, and laptop audio systems.
You use it from an AI assistant such as Codex, Claude Code, or Cursor on the same computer as REW. MCP is the connection that lets the assistant call these tools; this project is not a separate calibration app. REW does the measuring, and your DSP or receiver applies the changes it supports.
This project does not promise universally “perfect” sound. REW can measure an acoustic system; it cannot infer listener preference, microphone placement, or a speaker's safe output capability without evidence. The MCP reports missing data and keeps preference separate from engineering calculations.
Start here
New to MCP? Follow the written walkthrough. It covers what you need, how to install and connect the tools, and how to analyse your first REW measurement. You can begin with measurements you already have.
The MCP is free and open source. Your chosen assistant has its own account and billing requirements. Automated REW sweeps through the API require a REW Pro license; reading existing measurements does not require automated sweep control. See REW's API documentation.
Related MCP server: EvoBurrow MCP
Choose the right audio MCP
Use this project when you need | Use EvoBurrow MCP when you need |
General REW acoustic measurement and evidence-based EQ | A1 Evo AcoustiX artifact inspection or terminal workflows |
Room, speaker, subwoofer, car, powered-speaker, or laptop calibration | Denon/Marantz AVR status, preset mapping, backup, or guarded LAN control |
JamesDSP, CamillaDSP, Equalizer APO, FIR, or listening-test workflows | Audyssey |
The projects share measurement discipline, but this MCP deliberately stays hardware-neutral; EvoBurrow owns the A1 and AVR-specific safety boundary.
Measured laptop example — before and after
This is measured evidence from four repeated REW sweeps per state, averaged with 1/12-octave smoothing. Factory is amber and the cut-only JamesDSP preset is cyan. The shaded area below 500 Hz marks the protection-limited bass region where no boost was applied. The plot preserves measured SPL; preference was assessed in a separate level-matched listening comparison. Device, microphone, and preset identifiers are intentionally removed from the public artifact.
See the example explained for what this plot shows, what it does not establish, and how to compare changes on your own system.
Compatibility
Component | Support |
Node.js | 20 and 22 |
Operating systems | Windows x64, macOS Intel/Apple Silicon, Linux x64/ARM64 |
REW | Local API on port 4735; cross-platform install discovery and confirmed startup |
JamesDSP | Linux/JDSP4Linux inspect, preview, backup, apply, verify, rollback, blinded level-matched A/B |
Equalizer APO | Export; guarded apply when an explicit config path is configured |
CamillaDSP | YAML export; guarded file apply when explicitly configured |
miniDSP | REW-compatible text export; no hardware mutation in the beta |
A1 Evo/Denon | Optional separate bridge; no AVR assumptions in this plugin |
Install
Follow the installation guide for release downloads,
absolute file paths, and the configuration for your assistant. Codex uses
config.toml; the repository's .mcp.json is not automatically loaded by the
Codex CLI. The guide gives an explicit registration command.
For a source checkout:
npm ci --ignore-scripts
npm run build
npm test
npm run validate:releaseThe checked-in MCP configuration starts the self-contained dist/server.mjs.
The packaged CLI is also bundled, and the published artifact declares zero
runtime npm dependencies. Third-party build inputs are exact-pinned and installed
with lifecycle scripts disabled. Run
rew_install_discover to locate REW through a user override, the
AUDIO_REW_EXECUTABLE environment variable, PATH, and conventional Windows,
macOS, or Linux locations. If discovery fails, pass the absolute executable path
(or REW.app on macOS) to that tool or rew_launch_plan. Starting REW requires
the matching rew_launch_execute confirmation. Startup is verified against the
API on http://127.0.0.1:4735; override AUDIO_REW_URL only for a trusted local
network endpoint. REW's API still must be enabled in REW itself.
Guided workflow
The beginner walkthrough uses ordinary prompts and explains what to expect. The steps below are the tool sequence for users who want to work directly with the MCP interface.
Run
audio_doctor. If REW is offline, userew_install_discover, then the confirmedrew_launch_plan/rew_launch_executepair. Continue withrew_capability_negotiateandaudio_guided_session_plan.Inventory the host, REW, microphone calibration, output route, and DSP state.
Create a protected repeated-session plan. Audible execution requires fresh microphone-placement, area-clear, and route/safety confirmation.
Save separate raw traces, then run quality, dual-resolution, direct/late, crossover, distortion/compression, and human-listening analyses.
Prefer placement, polarity, timing, crossover, and stable cut-first EQ.
Apply only a hash-bound confirmed plan with backup and rollback.
Re-measure at matched level. Start
audio_post_eq_verification, then pollaudio_job_status; the verifier derives level difference from the traces. Predicted response and caller-claimed level matching are never acceptance evidence.Use randomized, level-matched A/B or ABX for preference/discrimination. On JamesDSP,
jamesdsp_ab_planplus its presentation and restore executors switch presets and host volume transactionally without revealing assignments.
Analysis views
Reports preserve the source trace and display native unsmoothed data, derived engineering resolution, 1/48-octave structure, adaptive modal-to-perceptual smoothing, and ERB/perceptual views. Filters are eligible only when features are stable across repetitions and held-out data. Narrow/spatial nulls are not boosted.
audio_report_plan can also accept two to four repeated comparison groups. It
averages each state, preserves ±1 SD, calculates the measured level difference,
and exports Markdown, HTML, JSON, and a standalone labeled SVG suitable for a
README. Missing SNR rejects the quality gate by default; exploratory callers must
explicitly opt out with requireSnr: false.
Privacy
There is no telemetry. Local sessions may contain usernames, absolute paths,
device names, room coordinates, microphone hashes, and preset fingerprints.
Never publish measurements/, sessions/, backups/, profiles/, reports/,
or filters/ directly. Use the redacted support-artifact tools and review the
result before sharing it.
Development and validation
npm test is hardware-independent. npm run coverage exercises deterministic
DSP and safety fixtures. npm run validate:release checks manifests, metadata,
the bundled server, and package contents. Real-system smoke tests are opt-in and
must never run in ordinary CI.
REW requests are bounded to four concurrent calls by default. Set
AUDIO_REW_MAX_CONCURRENCY from 1–8 only when the local REW instance has been
validated at that concurrency.
See SECURITY.md, CONTRIBUTING.md, and CHANGELOG.md. Licensed under the MIT License.
Measurement-science laboratory modules
Bounded tools cover GUM-style uncertainty budgets and Monte Carlo propagation, bootstrap repeatability, complex-transfer coherence and phase confidence, ISO-3382-aligned room screening, exploratory polar/directivity scans, clean-output ladders, held-out-seat complex multi-source optimization, regularized FIR proposals, controlled listening trials, immersive/SOFA metadata preflight, and reference-corpus auditing.
Screening, inspired, and standards-aligned never mean certified conformity.
ISO, IEC, ANSI/CTA, AES, and ITU conformity requires the complete current normative
document, prescribed setup and processing, calibrated instrumentation, and traceable
evidence. Optimization and FIR outputs remain proposals until protected, level-matched
hardware remeasurement accepts them.
Curated external datasets
audio_dataset_catalog exposes reviewed metadata for selected SADIE II, FLAIR,
MeshRIR, and RAVes artifacts without downloading them. Dataset acquisition uses a
hash-bound plan/execute pair, requires license acknowledgement, enforces an exact
byte ceiling, accepts only pinned Zenodo HTTPS URLs, refuses overwrites, checks the
upstream checksum, calculates SHA-256, and writes a provenance receipt under the
workspace datasets/ directory.
Remote entries and checksum-verified downloads are not automatically scientific validation. Readiness is calculated per domain and requires parsed, method-compatible references from at least two independent institutions with justified tolerances.
Laptop calibration
Laptop speakers remain fully supported, but this is the least general workflow and is intentionally documented last. Laptop mode starts at 120 Hz and -30 dBFS by default. Stop immediately on an unexpected route, silence, clipping, limiter activity, rattling, or distress.
Support the project
Audio Calibration MCP is free and open source. If it saves you time or improves your system, you can buy daredoole a coffee.
Available Tools
101 toolsaudio_artifact_createC
Create and validate a versioned, replayable calibration artifact from supplied evidence.
Inputs: session (required), sweeps, filters, analyses, +2 more.
audio_artifact_migrateB
Migrate a legacy calibration artifact to the current schema without writing files.
Inputs: artifact (required).
audio_artifact_validateC
Validate a versioned calibration artifact and report reproducibility gaps.
Inputs: artifact (required).
audio_capabilitiesC
Report platform, safety limits, REW endpoint, targets, and optional integration support.
Inputs: no named parameters.
audio_dataset_catalogA
List curated independent audio datasets with DOI, institution, license, measurement metadata, pinned artifact size, and checksum without downloading anything.
Inputs: no named parameters.
audio_dataset_download_executeA
Download one exact planned dataset artifact, enforce the byte ceiling, verify its pinned upstream checksum, calculate SHA-256, and write a provenance receipt without overwriting files.
Inputs: plan (required), confirmationToken (required), confirm, timeoutMs.
audio_dataset_download_planC
Create a hash-bound plan for one pinned licensed dataset artifact. No download occurs until the matching execute tool is explicitly confirmed.
Inputs: datasetId (required), maximumBytes (required), home, licenseAccepted.
audio_doctorB
Check Node, workspace, REW, host audio, JamesDSP, and configured DSP adapters without changing state.
Inputs: home.
audio_dsp_adapter_capabilitiesA
Inspect cross-platform DSP export/apply adapters and the explicit configuration needed for each.
Inputs: no named parameters.
audio_dsp_apply_executeC
Back up, apply, verify, and roll back an exact configured DSP filter-file plan.
Inputs: plan (required), confirmationToken (required), confirm.
audio_dsp_apply_planC
Preview and hash-bind an exact Equalizer APO or dedicated CamillaDSP filter-file change.
Inputs: adapter (required), filters (required).
audio_eq_design_planC
Create a hash-bound, cut-only EQ proposal trained on stable traces and checked against withheld traces.
Inputs: entries (required), deviceClass (required), maxQ, lowHz, +9 more.
audio_eq_proposalA
Create conservative cut-first parametric EQ suggestions from a live REW trace; this never applies filters.
Inputs: id (required), deviceClass (required), bands, lowHz, +3 more.
audio_evaluation_corpus_manifestC
Audit synthetic, external, loopback, and cross-tool reference artifacts for hashes, licenses, independence, provenance, and regression tolerances.
Inputs: external, crossTool, loopbacks, synthetic.
audio_evidence_registryB
List authoritative sources and exact claim boundaries used by the analysis engine.
Inputs: no named parameters.
audio_filter_export_executeC
Write and verify an exact cross-platform filter export.
Inputs: plan (required), confirmationToken (required), confirm.
audio_filter_export_planC
Create a hash-bound cross-platform filter export plan.
Inputs: file (required), format (required), filters (required), home.
audio_fir_designC
Design a regularized bounded mixed-phase FIR proposal and report causality, pre-ringing, latency, quantization, and deployment gates.
Inputs: targetResponse (required), measuredResponse (required), taps, maxBoostDb, +3 more.
audio_guided_session_executeC
Open an exact guided session record and return its safe next step; this does not emit audio.
Inputs: plan (required), confirmationToken (required), confirm.
audio_guided_session_planB
Create a hash-bound guided or expert calibration workflow without changing audio state.
Inputs: name (required), deviceClass (required), home, mode, +3 more.
audio_host_inventoryA
Read host audio devices and defaults without changing routing.
Inputs: no named parameters.
audio_job_cancelC
Cancel or suppress the result of an asynchronous analysis job.
Inputs: jobId (required), confirm.
audio_job_statusC
Poll an asynchronous analysis job without holding an MCP request open.
Inputs: jobId (required), includeResult.
audio_laboratory_listening_planC
Create blinded randomized MUSHRA- or BS.1116-inspired trials with hidden-reference and anchor readiness gates.
Inputs: method (required), systems (required), seed, trials, +2 more.
audio_laboratory_listening_reportC
Produce descriptive confidence and repeatability statistics for completed controlled listening trials.
Inputs: plan (required), responses (required).
audio_linked_stereo_eq_planC
Create regularized per-channel EQ with linked filter centers and bounded left/right gain differences; each channel must pass held-out validation independently.
Inputs: deviceClass (required), leftEntries (required), rightEntries (required), maxQ, +12 more.
audio_listening_test_planC
Create a fingerprinted, measured-level-matched randomized A/B or ABX listening-test plan.
Inputs: presetA (required), presetB (required), mode, seed, +7 more.
audio_listening_test_reportC
Summarize a completed listening test while keeping preference separate from objective quality.
Inputs: plan (required), responses (required).
audio_maximum_clean_outputC
Evaluate a protected level ladder for output compression, coherence loss, and limiting using AES75-inspired gates.
Inputs: levelRuns (required), minimumCoherence, maximumCompressionDb.
audio_multisource_optimizeC
Solve regularized complex multi-source weights across training seats and require held-out-seat evaluation before acceptance.
Inputs: targets (required), trainMatrices (required), maxGainDb, iterations, +2 more.
audio_multisource_physical_optimizeC
Coordinate-search bounded gain, delay, polarity, and fixed crossover controls across training seats with held-out verification.
Inputs: targets (required), constraints (required), frequenciesHz (required), trainMatrices (required), +2 more.
audio_polar_characterizationB
Analyze guided loudspeaker polar scans and produce exploratory directivity rows with a CTA-2034 claim boundary.
Inputs: angles (required).
audio_post_eq_verificationC
Accept or reject an EQ using separately measured before/after traces, matched state fingerprints, level match, quality gates, and repeatability.
Inputs: deviceClass (required), afterEntries (required), beforeEntries (required), levelMatchedWithinDb (required), +10 more.
audio_repeatability_bootstrapC
Estimate repeatability and a deterministic bootstrap confidence interval from repeated scalar measurements.
Inputs: values (required), seed, trials.
audio_report_executeC
Write and verify an exact human-readable and machine-readable calibration report set.
Inputs: plan (required), confirmationToken (required), confirm.
audio_report_planC
Create a hash-bound Markdown, HTML, and JSON report plan with optional overlaid raw, 1/48, ERB, and frequency-dependent curves.
Inputs: baseName (required), assessment (required), eq, home, +7 more.
audio_room_metricsC
Calculate EDT, T20/T30, C50/C80, D50, center time, decay linearity, and regression uncertainty from an impulse response.
Inputs: impulse (required), sampleRateHz (required), directIndex, minimumDecayRSquared.
audio_room_spatial_summaryC
Summarize position-to-position variance for room-acoustic metrics.
Inputs: positions (required).
audio_route_executeC
Apply an exact host routing plan, verify it, and return rollback values.
Inputs: plan (required), confirmationToken (required), confirm.
audio_route_planC
Create a hash-bound host audio routing plan. Native mutation is currently supported only through pactl on Linux.
Inputs: sink, source, sinkMuted, sourceMuted.
audio_session_advance_executeC
Advance one exact guided stage, preserving a recoverable session backup and returning the next tools.
Inputs: plan (required), confirmationToken (required), confirm.
audio_session_advance_planC
Bind completion of the current guided stage to concise evidence without changing the session.
Inputs: evidence (required), sessionFile (required), completedStage (required), home.
audio_session_replay_validateC
Validate whether a calibration artifact has enough immutable evidence for deterministic offline replay.
Inputs: artifact (required).
audio_session_statusC
Read a workspace-contained guided calibration session.
Inputs: sessionFile (required), home.
audio_sofa_metadata_assessmentB
Preflight HRTF/BRIR SOFA metadata before handing the HDF5 file to a maintained SOFA implementation.
Inputs: metadata (required).
audio_spatial_layout_assessmentC
Preflight immersive channel coordinates and head-position metadata without overstating BS.2051 compliance.
Inputs: channels (required), listener, headPositionMetadata.
audio_speaker_protection_assessmentC
Derive correction floor and permitted boost only from supplied capability, headroom, compression, and limiter evidence.
Inputs: headroomDb, measuredF3Hz, compressionDb, maximumBoostDb, +5 more.
audio_speech_transmission_screeningC
Calculate an IEC-60268-16-inspired speech-transmission screen from caller-supplied modulation transfer factors and importance weights.
Inputs: bands (required), redundancyPenalty.
audio_support_bundle_executeC
Write an exact redacted support artifact after rechecking its source hash.
Inputs: plan (required), confirmationToken (required), confirm.
audio_support_bundle_planC
Create a hash-bound plan for a redacted JSON support artifact; raw traces and identifying metadata are omitted.
Inputs: outputName (required), sourceFile (required), home.
audio_target_registryC
List versioned, evidence-labelled listening target starting points.
Inputs: deviceClass.
audio_transfer_qualityC
Evaluate complex transfer-function coherence, phase confidence, harmonic contamination, excitation, timing variance, and clock drift.
Inputs: bins (required), minimumCoherence, minimumInputPower, repeatTimingSeconds, +5 more.
audio_uncertainty_budgetC
Propagate calibrated measurement uncertainty using GUM-style linear combination and an explicit standards-claim boundary.
Inputs: components (required), coverageProbability.
audio_uncertainty_monte_carloC
Propagate normal, rectangular, and triangular input uncertainties with deterministic JCGM-101-style Monte Carlo sampling.
Inputs: components (required), seed, trials, estimate.
audio_workspace_scanC
Inventory profiles, sessions, measurements, backups, and reports in the AudioCalibration workspace.
Inputs: home.
car_channel_map_validateB
Validate a car-audio DSP channel map and flag missing protective crossovers, duplicate outputs, and invalid passbands.
Inputs: channels (required).
car_geometric_delay_advisorA
Calculate a geometry-only initial delay map for car speakers. Acoustic measurements supersede this estimate.
Inputs: listener (required), speakers (required), speedOfSoundMps.
jamesdsp_key_executeB
Back up JamesDSP, apply one exact key change, verify it, and automatically restore the active configuration on failure.
Inputs: plan (required), confirmationToken (required), confirm.
jamesdsp_key_planB
Create a hash-bound change for one JamesDSP key after verifying that the installed version exposes it.
Inputs: key (required), value (required), home.
jamesdsp_preset_executeB
Back up JamesDSP, execute an exact preset load/save plan, verify status, and automatically restore the active configuration on failure.
Inputs: plan (required), confirmationToken (required), confirm.
jamesdsp_preset_planC
Create a hash-bound JamesDSP preset operation with an automatic configuration backup.
Inputs: action (required), presetName (required), home.
jamesdsp_snapshotA
Fingerprint the live JamesDSP configuration and identify the exact active preset and effective EQ/bypass state without changing anything.
Inputs: no named parameters.
jamesdsp_statusB
Detect JDSP4Linux and decode engine, master-bypass, EQ-module, runtime-sync, and active-preset identity state.
Inputs: no named parameters.
rew_audio_configure_executeC
Apply and verify an exact REW audio configuration plan; restore the microphone calibration if REW clears it.
Inputs: plan (required), confirmationToken (required), confirm.
rew_audio_configure_planB
Create a hash-bound REW input/output configuration plan while preserving the current microphone calibration.
Inputs: input, driver, output, inputDevice, +4 more.
rew_audio_inventoryA
Inspect REW driver, devices, channels, sample rate, and microphone calibration without changing anything.
Inputs: no named parameters.
rew_capability_negotiateA
Detect the live REW API/version and supported read-only command surfaces before a workflow uses them.
Inputs: no named parameters.
rew_compression_analysisC
Analyze a matched REW sweep-level ladder for frequency-dependent output compression.
Inputs: entries (required), levelsDbfs (required).
rew_crossover_analysisA
Analyze phase agreement and predicted summation around a crossover. Verify the conclusion with a measured combined trace.
Inputs: subId (required), mainId (required), crossoverHz (required), spanOctaves.
rew_diagnostic_capabilitiesA
Read the live REW SPL, RTA, stepped-measurement, and generator command surfaces without running them.
Inputs: no named parameters.
rew_diagnostic_executeC
Execute an exact live REW diagnostic plan; audible operations require physical-readiness confirmation.
Inputs: plan (required), confirmationToken (required), confirm, audibleReady.
rew_diagnostic_planD
Bind an exact command exposed by the live REW diagnostic API.
Inputs: command (required), endpoint (required), parameters.
rew_direct_late_analysisC
Separate direct-window and later reflected impulse energy without treating the chosen gate as universal.
Inputs: id (required), ppo, lowHz, highHz, +4 more.
rew_dual_resolution_analysisC
Return raw/minimally smoothed and ERB-perceptual views so narrow engineering defects remain separate from broad listening interpretation.
Inputs: id (required), lowHz, highHz, stepErb, +3 more.
rew_generate_traceC
Compatibility alias: create an allowlisted hash-bound derived-trace plan; execute with rew_generate_trace_execute.
Inputs: processName (required), measurementIndices (required), parameters.
rew_generate_trace_executeC
Execute an exact derived-trace plan after rechecking the REW measurement set.
Inputs: plan (required), confirmationToken (required), confirm.
rew_generate_trace_planC
Create an allowlisted hash-bound REW derived-trace plan.
Inputs: processName (required), measurementIndices (required), parameters.
rew_human_listening_assessmentC
Start a cancellable asynchronous multidimensional listening assessment; poll audio_job_status for the result.
Inputs: entries (required), deviceClass (required), lowHz, snrDb, +8 more.
rew_input_level_checkA
Capture microphone input levels for a bounded period. This emits no sweep.
Inputs: confirm, durationMs.
rew_install_discoverB
Find a local REW installation across Windows, macOS, and Linux. An explicit absolute executable path can be supplied when automatic discovery fails.
Inputs: executablePath.
rew_launch_executeB
Start the exact discovered REW executable after explicit confirmation, then verify that the API becomes ready on port 4735.
Inputs: plan (required), confirmationToken (required), confirm.
rew_launch_planC
Create a hash-bound plan to start REW when its local API is offline. Uses automatic discovery or an explicit user-supplied executable path.
Inputs: executablePath, startupTimeoutSeconds.
rew_level_ladder_planD
Plan protected separately retained distortion/compression sweeps at several levels for one channel.
Inputs: role (required), saveFile (required), levelsDbfs (required), deviceClass (required), +9 more.
rew_load_fileB
Compatibility alias: create a hash-bound workspace MDAT load plan; execute with rew_load_file_execute.
Inputs: file (required), home.
rew_load_file_executeC
Load the exact hashed MDAT into REW after explicit confirmation.
Inputs: plan (required), confirmationToken (required), confirm.
rew_load_file_planC
Create a hash-bound plan to load a workspace-contained MDAT into REW.
Inputs: file (required), home.
rew_measurement_cancelC
Cancel a live REW measurement.
Inputs: confirm.
rew_measurement_executeC
Run an exact protected sweep plan, label and save results, then restore REW state. Audible confirmation and physical readiness flags are mandatory.
Inputs: plan (required), confirmationToken (required), confirm, areaClear, +1 more.
rew_measurement_planC
Create a guarded generic REW sweep plan for general speakers, cars, or laptops.
Inputs: saveFile (required), deviceClass (required), titlePrefix (required), outputChannels (required), +9 more.
rew_measurement_qualityC
Gate live REW traces on coverage, clipping, SNR, repeatability, route state, DSP state, and microphone-calibration identity.
Inputs: entries (required), lowHz, snrDb, highHz, +8 more.
rew_multiseat_analysisC
Analyze multi-seat consistency, modal candidates, seat outliers, and optional Schroeder frequency.
Inputs: ids (required), lowHz, highHz, rt60Seconds, +1 more.
rew_probeC
Check the local REW API and inventory live measurements.
Inputs: timeoutMs.
rew_repeated_session_planC
Plan 4-6 separately retained left, right, and combined traces with a complete route, volume, DSP, microphone, preset, and sweep fingerprint.
Inputs: channels (required), saveFile (required), deviceClass (required), titlePrefix (required), +9 more.
rew_save_allC
Compatibility alias: create a hash-bound Save-all plan; execute with rew_save_all_execute.
Inputs: saveFile (required), home, note.
rew_save_all_executeC
Execute and verify an exact REW Save-all plan without overwriting an existing artifact.
Inputs: plan (required), confirmationToken (required), confirm.
rew_save_all_planC
Create a hash-bound plan to save all live REW measurements into a new workspace MDAT.
Inputs: saveFile (required), home, note.
rew_traceB
Fetch a bounded REW magnitude/phase, group-delay, distortion, RT60, or impulse trace.
Inputs: id (required), kind (required), ppo, smoothing.
speaker_profile_getC
Read a workspace speaker profile.
Inputs: id (required), home.
speaker_profile_saveB
Deprecated compatibility alias: create a hash-bound speaker-profile save plan; execute it with speaker_profile_save_execute.
Inputs: profile (required), home.
speaker_profile_save_executeC
Save an exact speaker profile with pre-change backup, verification, and rollback.
Inputs: plan (required), confirmationToken (required), confirm.
speaker_profile_save_planC
Create a hash-bound evidence-tagged speaker-profile save/update plan.
Inputs: profile (required), home.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 101 tools
Most tools have clearly distinct purposes, especially with the plan/execute pairs and specialized analysis functions. Some aliases and related analysis tools (e.g., rew_trace vs. rew_generate_trace) create minor ambiguity, but overall an agent can differentiate the tools.
Naming is inconsistent: many tools follow the verb_noun pattern with _plan/_execute suffixes, but aliases (rew_load_file, rew_save_all) and direct-action tools (rew_trace, audio_room_metrics) break the convention. While prefixes (audio_, rew_, car_, jamesdsp_) are consistent, the mixed style is confusing.
With 101 tools, the server is massively over-scoped for any MCP purpose. Even for a complex domain like audio calibration, this number overwhelms agents and makes tool selection error-prone. The typical range is 3–15, and this is far beyond that.
The tool surface is extremely comprehensive, covering measurement, analysis, EQ design, DSP, listening tests, uncertainty, room acoustics, and REW integration. Minor gaps exist (e.g., no artifact deletion or explicit update for some resources), but the workflows are largely complete.
Maintenance
Related MCP Connectors
Check voice recording levels locally and share approved results with your assistant. OAuth required.
Audio features + harmonic set-building for tracks by name/ISRC. Spotify audio-features replacement.
Voice quality analysis: PVQA and AQuA MOS for WAV files, VoIP call analysis for PCAPs
Compose, mix, master and export in Logic Pro, REAPER or Ableton Live by talking to an AI engineer.
Related MCP Servers
- AlicenseAqualityCmaintenanceEnables natural language control of Room EQ Wizard for acoustic measurements, analysis, and EQ optimization, with peak/null detection and room-mode analysis.463MIT
- AlicenseNot gradedqualityBmaintenanceProvides a control room for inspecting A1 Evo AcoustiX artifacts, operating REW measurement workflows, coordinating calibration presets, and making allowlisted Denon/Marantz receiver changes with backups and verification.7MIT
- FlicenseCqualityBmaintenanceEnables reference-matching low-end mix analysis and adjustment in Ableton Live, including measurement, genre-scoped comparison, mix planning, safe apply/undo, and bass patch design toward a target.36-
- AlicenseAqualityBmaintenanceAudio perception for AI agents. Measure loudness, clipping, noise and pitch; view spectrograms; split audio into events; transcribe speech; compare before/after; and clean up, edit or synthesize sound. Runs locally, no API keys.14MIT