Acoustic Guitar Sustain Explorer

Quick start and user guide · v0.6.5

Observe how a guitar responds, how its sound changes after a pluck, and where notes behave differently across the fretboard. These four measurement views support the methods presented in Mastering the Sound of the Acoustic Guitar and in the workshops.

Sustain explorer shows how a note or chord changes through time. Tap frequency response measures the averaged response to repeated taps. Fretboard map records individual notes and compares their fundamental sustain, partial sustain and intonation. Live testing displays the current spectrum over a retained tap response.

The measurements help you locate a frequency region, compare instruments or adjustments, and connect what you hear with what you observe. They do not assign a quality score to the guitar.

Before you begin

Prepare the microphone and instrument

Connect the microphone and select Settings. Choose the Microphone and Capture channel. For a single-channel microphone, use channel 1.

Select Load calibration to open the microphone’s frequency-response calibration file, then check Apply microphone correction. Use the file corresponding to your microphone and its orientation. The correction file and its enabled state are remembered in the same browser when local storage is available.

Start requests microphone permission. Some browsers reveal device names only after this first request. In that case, start and stop once, return to Settings, and select the named microphone. Stop releases the microphone; the input-level meter is visible only during acquisition.

Work in a quiet room. Keep the guitar support, microphone position, input gain and playing method consistent. For tap measurements, use the positioning and tapping procedure described in Appendix A of the book Mastering the Sound of the Acoustic Guitar. For single-note measurements, damp the other strings without touching the vibrating string or unnecessarily constraining the soundboard.

Keep the measurement page visible during acquisition. Switching to another browser tab or hiding the page stops capture and marks it as interrupted.

Switch to full screen visualisation

Select ⛶ at the top right of the app to use the full screen. The enlargement button inside each graph enlarges only that graph within the application.

After a file operation that closes fullscreen, use Return to fullscreen when it appears, or the header’s ⛶ button. This changes the display, not the measurements.

Name the session

Enter the instrument or project in Project / guitar, for example “Lirica Walnut”. This name belongs to the complete session and is used for the Save all filename and PDF report.

Use Test · this tab to identify the current measurement or condition, such as “Open G chord”, “Body — strings damped” or “Original setup”. Each of the four tabs retains its own test name when you switch views. Individual exports include the project and test names.

A name is a label, not a new dataset. Renaming the fretboard test does not clear existing notes or separate measurements made before and after an adjustment. Save the session before starting a new condition; use New empty map for a separate map.

Enter the temperature in °C and relative humidity in % RH when available. These are manually entered session details, not sensor readings. Leave unknown values blank.

Under Settings → Session notes, record strings, microphone position, support, relevant adjustments and listening observations. The project name, environmental conditions and complete notes are included in saved sessions. The PDF includes the project details and a short excerpt of the session notes. Recording & notes → Observation notes provides additional notes for the current Sustain explorer recording.

Check the starting settings

For Tap frequency response and Live testing, begin with FFT length: 32k and Smoothing: 1/48 octave. These are the defaults. Their practical differences are explained below; they do not change the wavelet or fretboard sustain analysis.

The application may initially load the Lirica chord example. This is demonstration audio, not a measurement of the guitar named in your session. Record or load your own sound before preparing a report for another instrument. The example can be reopened through Settings → Recording & notes → Load Lirica example.

Microphone use requires a published HTTPS page. Recording, analysis, session saving and PDF generation run locally in the browser; the application does not upload your audio. The example recording, fonts and logo may load from website assets.

1. Sustain explorer

Record and inspect a note or chord

1. Capture one excitation. Select Sustain explorer and press Start. Leave approximately one second of quiet, then pluck one note or play one chord with a single stroke. Let it ring without repeating the excitation, speaking or moving the microphone. Press Stop after the sound has decayed sufficiently. The wavelet is calculated automatically.

2. Check the recording. The Recording graph shows the captured waveform. Use Listen to check the sound. To use an existing recording instead, select Load… and choose an uncompressed WAV. Include the attack, its sustain and preferably some quiet before the attack. For a multichannel WAV in this tab, select Settings → WAV channel after loading; changing the channel recalculates the analysis.

3. Select the useful interval when necessary. Hold the left mouse button and drag horizontally across the Recording graph, from before the attack to after the useful sustain. The application recalculates the wavelet for that interval while retaining the full original recording. Full recording removes the interval selection. To zoom the waveform without changing the analysed interval, use Shift + drag instead.

The orange marker identifies time zero, normally detected near the attack. Correct it with Shift + click on the Recording graph or through Settings → Recording & notes → Time zero. An interval selection changes the analysed segment; adjusting time zero changes the reference used for subsequent inspection and analysis.

4. Explore the result. Move the Time slider beneath the wavelet, or enter a time numerically. The vertical marker follows that time, and Spectrum shows the corresponding slice of the wavelet analysis. The mouse does not move this marker when hovering over the wavelet.

Read the wavelet

Time runs from left to right. Frequency runs from bottom to top on a logarithmic scale. Colour represents level. Follow the ridges to see which frequency components remain, which fade early, and where their level fluctuates.

The initial display covers 65 Hz–10 kHz, with a 50 dB colour range. One level reference applies across the map: frequency rows and time slices are not individually normalised. A ridge disappearing from view may simply have fallen below the displayed colour range.

Hatched areas near the recording boundaries indicate edge-affected analysis. Avoid relying on those areas when comparing the beginning or end of a component’s sustain. Include sufficient sound before and after the interval of interest.

Drag a rectangle on the wavelet to enlarge a region. Use Auto or double-click to fit the relevant content again. This changes the view, not the recording.

Inspect four partials of a single note

Choose Partials to display up to four components: P1 is the fundamental, followed by P2, P3 and P4. Check the detected f₀. Auto note repeats the estimate; entering f₀ manually lets you specify the fundamental when needed.

Each curve is referenced to its own early level, making their decay shapes easier to compare. Click a partial’s legend entry to emphasise its curve and frequency on the wavelet. Click it again to remove the emphasis.

For chords, use the wavelet and Spectrum. The partial view follows one harmonic series, not four separate notes in a chord. For numerical sustain results at individual string-and-fret positions, use Fretboard map.

Store and switch between measurements

After the analysis has finished, select Store measurement below the graphs. The app retains a snapshot of the recording, analysis, selected interval, time zero, microphone correction and observation notes. You can retain up to six stored Sustain measurements, separately from the six tap-response comparison curves.

Record or load another sound, then store it in the same way. Click a stored measurement’s name to display it again. Use ✎ to rename it, Ref to make it the comparison reference, and × to remove it from the stored list.

Store before replacing a measurement you need to keep. Recording, loading or selecting another measurement no longer opens the “Keep the current measurement?” prompt. An unstored working measurement can therefore be replaced without that reminder. Editing an existing stored measurement creates a working version; store that version to retain the changes as another snapshot.

At six stored measurements, save the session and remove an item before storing another. Removing an item from the list does not alter an already exported session file.

Compare two recordings

Choose Reference → Keep current as reference, or press Ref beside a stored measurement. Then record, load or select another measurement. Swap A / B switches between the current recording and its reference while retaining the same graph limits and colour scale. The spectrum view compares the two recordings at the selected time after their respective time-zero markers.

The wavelet uses a common level reference across the current recording, comparison reference and stored Sustain measurements. Individual recordings are not separately normalised when switching between them. Keep microphone position, gain, excitation and graph limits consistent.

Store, Reference and Save serve different purposes: Store retains a measurement in the current workspace; Reference selects an A/B comparison; Export → Save this tab · session or Save all tabs · session writes a reloadable file. Save before closing the page.

2. Tap frequency response

Measure the body response

1. Prepare the measurement. Damp the strings and use the support, microphone position and tapping method described in Appendix A. Select Tap frequency response, name the measurement and check the microphone correction.

2. Start the continuous analysis. Press Start and begin a consistent sequence of taps. The spectrum updates automatically. There is no separate recording to window and no Calculate FFT step.

3. Let the average fill. Watch the counter reach 32 / 32 averages, then continue tapping until the response is reasonably stable. These are FFT frames, not 32 individual taps. The analyser continuously replaces older frames with newer ones; it does not stop automatically when the counter reaches 32.

4. Retain the response. Press Stop while the tapping sequence is still represented in the average, rather than leaving a long silence first. The current tap response is retained automatically for Live testing and can be linked to the Fretboard map. Reset averaging, available during acquisition, clears the averaging history without deleting stored comparison curves.

The graph initially covers 65–1500 Hz on a logarithmic axis with a 50 dB vertical span. Move the pointer across it to read frequency and the current trace’s level in the large display above the graph.

Store and overlay several curves

After stopping, select Store curve to preserve a comparison copy. You can keep up to six stored curves, in addition to the current response. Give each a distinct name using its pencil button. Its checkbox shows or hides it independently; × removes that stored curve.

Measure the next condition and store it in the same way. Stored curves also appear in Live testing. Hiding a curve does not delete it, and starting another measurement does not overwrite these stored copies.

Use Export → Save this tab · session or Save all tabs · session to keep them after closing the page. Stored curves are not automatically saved to disk.

FFT length, smoothing and averaging

Open Settings → Tap frequency response & live testing.

32k provides a shorter observation and faster response. 64k uses twice as much recorded time per frame and halves the unpadded frequency-bin spacing. At a 48 kHz sample rate, the observations are approximately 0.683 s / 1.46 Hz for 32k and 1.365 s / 0.73 Hz for 64k. The finer plotted grid does not, by itself, provide additional resolving power.

Choose None when inspecting the detailed shape of narrow peaks; choose 1/48 octave for a lightly smoothed comparison. Smoothing changes the displayed spectrum, not the underlying stored spectral values.

Change FFT length while stopped. It applies to the next acquisition; it does not recalculate previously stored curves at a new resolution. The settings line beneath the graph identifies the current result’s FFT size and indicates a different selection for the next acquisition when applicable.

Averaging combines linear power spectra before conversion to decibels. It is separate from frequency smoothing. Neither control changes the wavelet bandwidth or the Fretboard map’s note-tracking routine.

3. Fretboard map

Measure notes without selecting every fret

The Sustain / Intonation buttons select two analyses of the same note recordings. There is no separate intonation capture process. Begin in Sustain to inspect the note-acquisition results.

1. Select the string. Choose Fretboard map, name the map and select the string using the String menu or its label at the left of the fretboard. The map covers six strings in standard tuning, from the open string through fret 14: 90 positions. You can measure only the region of interest; completing every position is not required.

For intonation as well as sustain, start each string with an open-string recording and repeat it at the end. Tune before the sequence, not during it. The detailed intonation procedure follows below.

2. Measure the background. Press Start and keep the strings quiet while the central display reads Measuring background. Wait for Ready — play the next note. The initial background measurement takes approximately 1.2 seconds.

3. Play once and let the note ring. Pluck one note on the selected string. The app identifies the pitch, highlights the corresponding fret and captures the sustain. The display shows the detected note, string, fret and elapsed time. Do not play another note while it is still capturing or analysing.

4. Follow the Ready prompt. After Analysing note and Note stored, wait for Ready — play the next note. Damp any remaining sound before the next pluck. Continue along the string, skip positions or repeat a note. Repetitions become separate takes. The application retains the initial background estimate rather than requiring a new long silence after every note.

You may change the selected string while the app is Ready. Otherwise, press Stop, choose another string, and start again. Stop finishes the listening session and releases the microphone.

5. Review exceptions after the sequence. Click a position to inspect its results. Use the take selector to review repetitions or a Check entry for an unassigned recording. Use Listen to check what was captured. When you know the fret but the app assigned it incorrectly, select Correct fret, enter a value from 0 to 14, and let the app reanalyse that take. The correction applies to its recorded string; it does not repair a recording made with the wrong string selected.

Correcting the fret establishes where to analyse. It does not force a valid sustain result from weak, interrupted or unresolved sound. Delete take removes only the selected recording.

How long should each note ring?

The app follows the partials and finishes the capture automatically when they have decayed sufficiently for the acquisition’s stopping rule. In this revision, the background is also evaluated within the partial-tracking bands. A trackable partial prevents the overall recording level from declaring the note quiet prematurely. A short tail is retained after the partial-range endpoint.

Each note still has a 20-second acquisition limit. Stop remains available, and difficult notes can be retained for checking rather than assigned confidently. A genuinely persistent note can require a longer capture than in earlier versions. An automatic stop does not guarantee that every partial will pass the subsequent sustain-quality checks.

Let the app finish whenever practical. Stopping or damping a note early can leave an incomplete decay interval or an irregular envelope. Such a result may be reported as a lower bound, insufficient range or a complex result rather than a numerical sustain time.

Loading an older session does not recover sound that was never recorded. When an earlier capture stopped before the full decay interval, repeat that note to obtain a complete measurement; reanalysis cannot reconstruct its missing tail.

For an existing single-note recording, select its string, choose Load…, and open the WAV. Include quiet before the attack and keep the complete file below 23 seconds. In this revision, Fretboard map imports the first WAV channel. Use a mono file when the required signal is on another channel.

Identify frequencies directly on the fretboard

In Sustain view, select P1, P2, P3 or P4, then press Hz. Each cell displays the nominal frequency of that partial, while its sustain colour and measurement flags remain visible. Press Hz again to return to sustain times.

These are reference frequencies for standard tuning at A4 = 440 Hz, not measured peak frequencies. They are available even at unmeasured positions. Read the actual recorded frequency, including a detected peak pair, in Partial results or Peak structure.

This makes it possible to locate positions whose fundamental—or a higher partial—lies near a frequency of interest on the body response.

Check intonation using the same recordings

1. Establish the open-string reference. Tune the selected string before measuring it. Press Start, wait for Ready, then pluck the open string once and let the app finish. Intonation needs a usable measurement of the open string’s fundamental, P1; a correctly identified note name alone is not sufficient.

2. Measure the fretted notes. Continue with the usual acquisition process through the positions of interest, up to fret 14. Keep the plucking method and fretting pressure consistent, and avoid intentional bends or vibrato. You do not need to measure every fret. Do not retune during that string’s sequence.

3. Repeat the open string. Record it again at the end, without retuning first. This checks whether the reference remained consistent. One usable open-string take is enough for a provisional chart, but the app then reports One open reference; drift not checked.

The reference is the median of the usable open-string P1 measurements. If they span more than 4 cents, the app withholds that string’s relative intonation values and asks you to check the reference. The disagreement may reflect tuning drift or inconsistent measurements; it does not prove which occurred. Inspect the open-string takes. When a new tuning condition is needed, save the work and begin a separate map rather than combining references from different tuning conditions.

4. Select Intonation. After stopping acquisition, press Intonation above the fretboard. Each measured position shows its deviation from equal temperament, relative to that string’s measured open P1. Older saved recordings are analysed on first use; wait for Analysing intonation to finish. Their stored sustain results are not changed by this calculation.

5. Read the chart and inspect a position. The chart plots cents against fret number. Its string checkboxes show or hide each string independently. Click a point or fret to select the position. Numerical cells show the median of valid repeated takes; the vertical range markers show the spread between those takes, not a calibrated uncertainty interval. A gap remains where no reliable value is available.

The detail line distinguishes the selected take’s measured P1, its variation within the measurement interval, and the map median for that position. It also shows the measured open-string reference and its consistency status. Return to Sustain to choose another repetition with the take selector, correct a fret or inspect the partial curves and signal-quality table.

Read the intonation values

Positive values are sharp; negative values are flat. The open string displays 0 ref, because it defines the reference. This does not mean the open string is exactly in tune with A4 = 440 Hz.

Blue indicates a value below −2 cents, orange above +2 cents, and the neutral colour covers −2 to +2 cents. These are directional display bands, not validated acceptance tolerances. The red–orange–green sustain thresholds do not apply to this view. The P1–P4 and Hz controls are hidden while Intonation is selected.

— means unmeasured. ? means that no reliable relative value is available, either for the note or for its open-string reference. It is not zero cents. Inspect the explanation below the chart.

For fret n, the calculation is:

cents = 1200 × log2(measured P1 / measured open-string P1) − 100 × n

The app measures P1 directly from the recording. Note recognition and Correct fret identify where to look; a nominal frequency supplied by fret correction is not treated as a measured frequency. The intonation calculation does not average P1–P4 or substitute a higher partial for an unmeasurable P1.

Every note uses the same 0.15–0.80 s after the detected attack interval. Three overlapping 350 ms Hann windows provide frequency estimates, combined by a median. The checks require an independently estimated noise floor, at least 25 dB of estimated P1 signal-to-noise margin, and no more than 6 cents of variation between the three windows. Weak or competing P1 peaks, unresolved splitting, excessive pitch variation or a compromised recording can prevent a result even when the fret was identified correctly.

Sustain and intonation have separate validity checks. A note marked > lower bound for sustain may still have valid intonation, because the early pitch interval is present even though the complete sustain interval is not. Conversely, a measurable sustain envelope does not guarantee one sufficiently stable P1 frequency for intonation.

The displayed decimal places are not an accuracy specification. This is an intonation comparison, not an automatic nut- or saddle-compensation calculator. Use Export → Intonation results · CSV for the numerical results and per-take evidence; save the session to retain the audio and continue working.

4. Live testing

Compare played notes with the retained tap response

Select Live testing after measuring a tap response, then press Start. Play individual notes or carry out the local checks described in Appendix A.

The live trace uses the same FFT length, Hann window, overlap and smoothing as Tap frequency response, but only four averages. The retained tap response remains underneath, and stored comparison curves can be shown or hidden using their checkboxes.

Move the pointer across the graph to read frequency and current-trace level above it. Use the tap curve to locate body-response features, then observe where the played note’s components fall. Compare frequency positions without assuming that a pluck and a tap have the same excitation level.

Peak displays the maximum levels reached by the live averaged spectrum during the acquisition. Turn it off to show the changing trace without that additional curve. Reset averaging clears the current averaging history and peak history. Stop freezes the live result without changing the retained tap response.

The default live view uses a 100 dB vertical span, initially −120 to −20 dBFS when uncalibrated. Auto and Limits let you adapt the view to the signals being compared.

Understanding the fretboard results

What each sustain cell measures

In Sustain view, the map shows the selected partial’s 20 dB sustain time in seconds. P1, P2, P3 and P4 are measured independently: the software does not average the fundamental and overtones together.

When a position contains repeated takes, its numerical cell value is the median of valid results for that same partial, string and fret. The graphs and Partial results table below describe the selected take, which may differ from the map median.

The sustain number comes from a fitted decay slope over the observed −5 to −25 dB part of that partial’s envelope. It is not the total audible duration of the note, a reverberation time or an extrapolated 60 dB decay. An incomplete or unreliable interval is not replaced by an invented number.

Colours and measurement states

The legend below the fretboard covers these states. A question mark does not necessarily mean the note was misidentified: pitch recognition may be reliable while its sustain cannot be measured reliably.

To change the colour limits, select the partial first, then open Settings → Fretboard map and edit Red below and Green from. Each partial has its own limits. Changing them recolours the map without changing measurements. They are comparison limits, not validated wolf-note thresholds or a grading system.

There is no automatic frequency compensation of the sustain values or colour limits. The same limits for a selected partial apply across the fretboard. A chord’s higher-frequency ridges are not used as an expected-decay curve for every string and fret. Compare nearby notes and their measured envelopes before treating a shorter sustain at a higher pitch as a local problem.

Inspect the selected note

Partial results gives each partial’s frequency or detected peak pair, its Start dB and its sustain result. Start dB expresses the early level relative to the strongest of the four measured partials; it is not absolute SPL.

The four decay curves each use their own early reference, allowing comparison of their decay shapes. Click a table row or legend entry to emphasise a partial. The Map readout gives that partial’s median and valid-take count; Repeatability reports the spread between repeated valid measurements.

Low signal means that a dependable result is unavailable at the recorded level. Too short indicates insufficient decay range, not necessarily an intrinsically short-sustaining note. Too fast means the decay is too rapid for a reliable estimate with the analysis used. Complex identifies an envelope poorly represented by one slope. Not separated means the partial could not be isolated reliably from neighbouring frequency content.

These are different outcomes. In particular, a strong nearby body resonance can interfere with a partial even when the note itself is correctly identified. Inspect the curves and repeat the measurement rather than treating every unavailable number as a note-recognition failure.

Examine peak structure and the body response

Peak structure shows an unsmoothed close-up spectrum of the selected partial. When two peaks are detected, their frequencies and separation are displayed. Doublet indicates that the pair also passed the check in earlier and later analysis windows; Two peaks — inspect remains a more provisional observation.

A detected pair is not automatically a wolf note, and it does not establish a structural cause. Compare the peak structure, decay envelope, repeated takes and what you hear. Enlarging the graph cannot recover detail absent from a short or weak recording.

Choose Body response to see the partial frequencies over the linked tap curve. Selecting P2 highlights the second partial’s frequency, not the fundamental. Use Use retained tap curve to link or replace the body-response reference with the latest tap measurement.

An available tap response is linked when entering an empty map. An existing map keeps its linked response until you replace or clear it. Check the curve name before comparing instruments or build conditions. The app does not assign automatic Air, Top or Back labels.

Follow the harmonic balance through time

Move Time beneath the selected-note graph. The four bars show the P1–P4 levels at that instant against one fixed reference: the strongest initial partial. They are not renormalised at each time.

This helps distinguish a component that starts weakly from one that starts strongly and fades quickly. A partial too close to the noise floor is left unavailable. Successful pitch recognition does not guarantee that the fundamental itself is strong enough for a valid sustain measurement.

Check signal quality

In Sustain view, read Signal quality for the selected take and partial. It reports input peak and clipping, the partial’s signal-to-noise margin, pitch assignment, decay status, repeatability and the reported browser-processing status. When a lower bound is stored, the decay-range line identifies the incomplete range and whether that capture ended on a quiet stop.

Avoid clipping and repeat affected takes with lower input gain. Keep the microphone close enough to obtain a clear signal while preserving the chosen measurement position. Reduce background noise rather than assuming that increasing gain will improve the signal-to-noise ratio.

The sustain acceptance rules require approximately 35 dB of initial signal-to-noise margin in the partial’s analysis band, with at least 10 dB remaining above the estimated floor at the lower decay endpoint. A weak fourth partial does not invalidate a measurable fundamental. Dotted parts of the curves identify low-level regions with reduced reliability.

The app requests automatic gain control, noise suppression and echo cancellation off. Reported off describes the browser’s report, not independent verification of the whole microphone and operating-system path. Imported recordings remain unverified. Beating or a changing decay slope can be properties of the note rather than recording faults; their curves remain useful even without a single sustain number.

Levels: dBFS and optional calibrated SPL

Without absolute calibration, the spectrum graphs display dBFS, referenced to the digital signal’s full scale. Loading a microphone frequency-response file does not by itself establish absolute SPL. The UMIK sensitivity header alone does not verify the browser and operating-system gain.

To establish an absolute reference, use Live testing with a steady sound whose level at the microphone is independently known. While the live analyser is running, open Settings → Level calibration, enter Known steady sound level (dB SPL), then select Match current live level. Alternatively, enter an already verified value for SPL at digital full scale. Show calibrated SPL enables the calibrated display.

The default “94” in the reference field is only an entry value; it is not evidence that the microphone has been calibrated. Calibrate before collecting the curves you intend to compare, and keep the microphone, capture channel and input gain unchanged.

Stored spectra retain their own level calibration; a new reference does not retroactively calibrate earlier curves. When visible curves do not all have an absolute reference, their shared graph uses dBFS rather than mixing units. The live input meter continues to show digital level for clipping checks. Sustain times and relative partial levels do not require absolute SPL calibration.

Graph controls

In Tap frequency response and Live testing, move to the upper-left area of the graph to reveal its tool strip. On other graphs, hovering reveals the controls.

In the fretboard detail graphs, Auto includes the displayed curves, their low-level dotted portions, fitted-line endpoints and any displayed comparison. The body-response view also includes the selected note’s upper-partial frequencies. Choosing another note, take or partial clears the previous detail-graph limits. Auto on the intonation chart includes the visible strings and their repeated-take ranges.

With a graph focused, A runs Auto, 0 resets, Enter opens Limits, and Ctrl/Cmd + Z undoes a graph-view change.

Use the Time slider to move the wavelet’s inspection marker. Shift + click on the Recording graph changes time zero; on a partial-decay graph it moves the time cursor. Ordinary graph zooming and panning do not alter measured data. Selecting a Recording interval or adjusting time zero does recalculate the associated analysis.

Save, reopen and export a session

Save work for later

Stop acquisition and wait for analysis to finish, then open Export.

Save this tab · session saves the active tab’s data and associated references. In Sustain explorer, this includes its stored measurements. In Fretboard map, it includes the note recordings, sustain results and available intonation analysis.

Save all tabs · session saves the complete workspace: the current and stored Sustain measurements and comparison reference, current tap and live spectra, stored tap curves, fretboard takes—including unassigned recordings—linked body response, available intonation results, calibration, settings, project and test names, notes and graph views.

The complete-session filename is based on Project / guitar, for example Lirica_Walnut_all_tabs.sustain.zip, not on the active tab. A single-tab export includes the project name, that tab’s test name and its tab identifier.

These commands produce a local .sustain.zip file. Keep it intact. Use Export → Load saved session… or Load… to reopen it. Loading replaces the tabs contained in that file and restores its shared settings and session details. Save your current work before replacing it.

Sessions retain raw audio where the app records it: the Sustain explorer capture and individual fretboard notes. Tap frequency response and Live testing retain spectra, not a continuous audio recording of the acquisition.

The Unsaved changes / No unsaved changes indicator tracks changes relative to session-file exports and loads. Store measurement, Store curve, PDF and CSV exports do not replace a saved session. After saving, check that the file is present in Downloads or your chosen folder.

Measurements are not automatically saved when the page closes. Calibration and preference storage are not a substitute for a saved session. Large maps can reach the 256 MiB note-audio limit or 780-take limit; save before starting a new map.

Keep a map comparison reference

In Fretboard map, Reference → Keep current as reference retains a numerical snapshot. When you inspect a position, the selected partial’s previous curve and median can be compared with the current result. This snapshot does not duplicate the audio.

Use separate saved sessions for different guitars or build conditions. Settings → Fretboard map → New empty map clears the current map, its linked body curve and its comparison reference; save first.

Reanalyse saved takes processes the retained note audio again. Save a copy beforehand to preserve the previous results. Changing FFT length or spectrum smoothing does not require reanalysing the map.

Export an A4 portrait session summary

Select Export → Session summary · PDF (max. 2 A4 pages). The application creates and downloads the PDF directly, using the project name for its filename.

Page 1 — Acoustic session. Project name, temperature, humidity, input information and measurement counts, followed by the displayed Sustain wavelet, tap response and live-testing result. A short excerpt of the session notes is included when available. The report uses the chosen wavelet; it does not add a page for each stored Sustain measurement.

Page 2 — Fretboard sustain and intonation. Four maps show the median valid sustain results for P1, P2, P3 and P4, from the open strings through fret 14. The intonation chart and open-string reference information appear below. All four sustain maps are included regardless of the partial currently selected in the app.

The report is limited to two A4 portrait pages. With no fretboard recordings, it contains only the first page. Missing or unreliable measurements remain unavailable. Individual take tables, unassigned recordings, detailed peak inspections and full notes are not expanded into extra pages; use the numerical exports and saved session for those details.

Before exporting, check Project / guitar, the individual test names, temperature, humidity and notes. In Sustain explorer, display the recording intended for the report rather than the Lirica example or an unintended reference. Check which tap comparison curves are visible. The report includes all six strings’ available intonation results rather than using the on-screen string-visibility selection.

When necessary, export first calculates intonation from the existing note recordings. It does not record new sound or alter their stored sustain results. Wait for this calculation and PDF generation to finish.

The PDF is for reading and sharing. Save the session file as well to reopen recordings and continue working. Exporting the PDF alone does not save a reloadable workspace.

Export individual results

The Export menu provides Graph image · PNG, Spectrum · CSV, Four partials · CSV, Map results · CSV, Intonation results · CSV, Original recording · WAV and Measurement data · JSON where applicable.

Map results · CSV contains per-take partial frequencies, levels, sustain results, lower bounds and measurement states rather than only the coloured-cell medians. Intonation results · CSV contains the position medians and a separate section of per-take P1 estimates, reference-relative cents, within-note variation and validity information. A missing reference can leave relative cents unavailable even when a take has a usable P1 estimate.

Unassigned recordings are retained in the full session; the PDF reports their count rather than listing each one. They do not appear as assigned fret positions in either map CSV. In the Fretboard map tab, Graph image · PNG exports the currently selected Sustain or Intonation view.

WAV exports contain uncorrected, non-normalised audio. In Sustain explorer, Original recording · WAV exports the full original selected channel, even when a shorter analysis interval is selected. In Fretboard map, it exports the selected note capture, including its pre-attack buffer—not the entire listening session.

PNG, PDF, CSV and measurement JSON exports do not replace a complete reloadable session file.

Keeping useful comparisons

Record the strings, support, microphone position, gain and relevant adjustments. Compare measurements made under consistent conditions, then return to listening. The app records sound at the microphone; it does not directly measure bridge mobility or the energy remaining in the strings.

A short sustain, a split peak or proximity to a body resonance is a frequency region to investigate—not, on its own, an instruction to change the instrument. Repeat uncertain notes, compare their partials and inspect the evidence before drawing conclusions. The numerical checks do not establish universal diagnostic criteria for real guitars.