A setup and workflow guide for the measurement tools on this website. Choose your equipment, capture or import real audio, understand the result and carry suitable measurements into crossover design.
The page now guides you through Choose → Set up → Results. Select a goal first, follow its checklist, then save your result. Optional settings stay collapsed until needed.
Set up the speaker before you measure
Put the speaker on stable support with clear space between it, the microphone and nearby reflecting surfaces. Use the baffle or cabinet you intend to design around. A whole-speaker check keeps its protective crossover connected; an individual driver needs a suitable sweep band and protection.
Place the microphone at your chosen reference axis on a stand. Record distance, height and orientation, following its calibration-file orientation. For Phone Quick measure, rest the phone steadily with its microphone opening clear; its initial placement suggestion is 0.5–1 m, for broad comparisons. Keep this geometry, input gain and playback level fixed when repeating or comparing.
Connect the playback source’s line output to the amplifier and the amplifier to the loudspeaker. Connect the microphone separately through USB or a microphone interface. Select and verify the input on the page. Start at low amplifier volume and check for clipping. Use the external level reference only if you have a suitable meter; otherwise choose relative dB.
Capture twice, inspect the raw impulse, identify the direct arrival and choose a window that ends before the first reflection. More clear space may allow a longer reflection-free window; no fixed placement guarantees a particular gate or bass accuracy.
Pair: put PC and phone on the same Wi-Fi. Create the QR code on the PC and scan it with the phone camera. If automatic pairing is unavailable, the phone shows a reply. Share that text to the PC, paste into Reply from your phone and click Connect phone.
Position: rest the phone 0.5–1 m in front of the speaker with its mic opening clear. Enter an optional name, tap Allow microphone, then Ready to measure. Check the input’s processing status on the PC and confirm that it is your phone. Actual microphone selection is controlled by the phone/browser; attaching an external mic does not prove the browser chose it.
Record: start with low amplifier volume. The default 3-second sweep covers 300 Hz–10 kHz. Tap Run sweep on the PC; the phone starts recording before playback. Two short timing sweeps bracket the main sweep, allowing an estimated clock correction. Keep both browser pages visible until the lossless audio arrives on the PC.
Interpret: the first graph is unsmoothed magnitude. Save it as Before, change one thing, and repeat for After. Keep position, orientation, input gain and amplifier volume fixed. Changed settings trigger a comparison warning. Smoothing and graph limits help reading; they do not improve microphone accuracy.
Export: save current frequency/magnitude as FRD or CSV, or save unchanged original 32-bit float WAV. Phase is omitted because the two devices do not share an electrical timing reference. A short optional gate reduces reflections but removes low-frequency information; a 6 ms window exports from about 333 Hz upward. Use calibrated microphone/loopback data for precision crossover summation.
Optional level reference
After pairing and confirming the phone, expand Optional: match an external level reading. Check low volume, then Play reference white noise. Note a steady external Z / unweighted reading with both sensors together. Enter that reading and Capture level reference without moving or changing gain. All three processing controls must report off. Matching is limited to the current phone, source and sweep settings. Repeat after a gain/device change; software cannot detect every external gain adjustment. This is an estimated level reference, not phone frequency calibration or meter certification.
When it does not connect or measure
Use the same normal Wi-Fi; guest-network isolation and firewalls can block a direct connection. QR links expire here after 10 minutes. Recreate the link if it is stale. Phone processing may be reported on or unknown; treat those curves as broad comparisons only. Clipping, insufficient signal above background, missing timing sounds, clock disagreement and incomplete recordings produce a retry instruction. Locking the phone or changing apps stops capture. End session stops microphone and playback. The usual external-source workflow remains available if direct pairing fails.
Audio is sent directly between your devices, not to a SpeakerFoundry audio server. Results live in the current PC tab until saved or closed. Manual pairing links include connection details; share privately. Real phone/hardware acoustic validation and verified model-correction profiles remain future work.
Open a recording or response → Frequency spectrum (quick look)
A supported audio recording
One-channel spectrum; CSV export
Prepare measured driver data
Open a recording or response → Driver response + phase (stereo transfer)
Simultaneous microphone and loopback recording, shared reference and appropriate calibration
Transfer magnitude, phase and coherence; CSV or qualified FRD
Make a test signal
Create a test sound
Suitable playback equipment and a safe driver frequency band
Logarithmic sweep playback or WAV
For every workflow, check the input, begin at a low playback level, avoid clipping and keep your setup repeatable. Use an external calibrated microphone for accurate acoustic work.
01 / PHONE OR COMPUTER
A quick relative spectrum
Open this page over HTTPS in a supported browser. Place the phone or microphone at a repeatable position facing the loudspeaker. Keep the microphone unobstructed.
Open Use my microphone, tap Start microphone and allow microphone access. After permission, available microphones appear in the selector. Stop, choose an input and start again to change it.
Play sound through the loudspeaker using a separate player. Keep volume low initially. The meter shows input peak and RMS; reduce input or playback level if it reports clipping.
Read the spectrum as relative level versus frequency. Built-in phone processing can affect it; it is not an isolated driver response or an absolute SPL measurement.
Use Save results (CSV) to save the visible data. Tap Stop microphone on the Results screen when finished. Backgrounding the page also stops capture.
Use a calibrated microphone, a two-input audio interface with a shared recording clock, suitable recording software, an amplifier and the driver installed on its intended baffle.
Record the microphone and loopback reference simultaneously as one stereo file. Use a suitable broadband excitation signal for H1 transfer analysis. This page analyses the recording; it does not make that stereo recording itself.
Keep reference level, recording clock, microphone position and timing consistent when measuring each driver. Preserve polarity and do not independently time-align or normalise files used for crossover summation.
Open Open a recording or response, choose the stereo file (up to 30 MB / 120 seconds), select Driver response + phase (stereo transfer) and identify which channel contains the microphone.
Click Analyse audio. Check the frequency band and coherence. Low coherence or a weak reference calls for a better recording; the export retains only bins with coherence ≥ 0.80 and adequate reference excitation.
Apply the correct calibration, confirm the common-reference checkbox in Results, then Save driver response (FRD). Import each driver's FRD in the measured crossover workspace. Impedance needs a separately measured ZMA file; this tool does not measure impedance.
The loopback must use the interface's permitted line-level input range. Never connect a speaker-level amplifier output directly to an audio interface or phone input.
Open Create a test sound. Choose duration, start/end frequencies and digital peak level. Use a frequency range suitable for the connected driver.
Select the output device in your phone/computer sound settings. Browser playback follows that system output; a phone's own speaker cannot measure your loudspeaker.
Start with low amplifier volume and confirm the playback checkbox. Click Play sweep, or Download sweep WAV for an external player. The stop button ends playback.
This test-sound route generates the signal only. Choose Measure a gated driver response to record the microphone and digital reference together, recover an impulse and apply a time window. Room decay and RT60 are not included. A separately recorded sweep without a simultaneous reference is not a reliable acoustic phase file here.
Keep a tweeter's protective crossover in place for low-frequency tests. Digital level in dBFS does not specify loudspeaker power or acoustic SPL.
Frequency correction: load the microphone's verified response-error file with frequency in Hz and error in dB, using the correct sign convention. This tool subtracts that error only within the file's frequency range. Clear the file if you switch microphones.
Level offset: enter an offset established by an appropriate reference calibration. With the default 0 dB offset, microphone and single-channel results remain dBFS; entering an arbitrary offset does not make them calibrated SPL. Stereo transfer is a measurement/reference ratio until calibrated appropriately.
Changes: changing a recording, channel or calibration clears the previous result. Capture or analyse again before exporting. FRD export requires stereo transfer, valid phase and your common-reference confirmation; spectra export as CSV only.
Inspect an existing audio recording
Open Open a recording or response and choose your file. WAV is suitable; other formats depend on your browser. The limit is 30 MB and 120 seconds.
Select Frequency spectrum (quick look). Open Recording channel to choose left/channel 1 or right/channel 2. A mono file uses its available channel.
Click Analyse audio. Read frequency in Hz along the logarithmic horizontal axis and level along the vertical axis. With a 0 dB offset and no correction, the level is referenced to digital full scale.
Use the peak indication to check for clipping. The graph shows the recording's frequency content, which includes the source signal, loudspeaker, microphone and room. It does not isolate driver response.
Save the data using Save results (CSV). To use acoustic phase in crossover design, follow the stereo transfer workflow instead.
The analyser uses an 8192-sample Hann window and 50% overlap for recordings. It averages up to the first 120 complete windows, so it does not necessarily analyse every second of a long recording.
Bring REW measurements into SpeakerFoundry
REW .mdat files are saved measurement sessions. SpeakerFoundry imports their exported response text; it does not open the native session, recover its impulse response or apply REW's measurement settings.
Open your .mdat session in REW. Choose File → Export → Export measurement as text, or export all measurements as separate files.
Export frequency in Hz, level in dB and measured phase in degrees. Use a decimal point, no digit grouping, and space, tab or comma separators. Keep comparable level, timing and window settings across drivers; do not independently normalise their sensitivity.
Each response must contain 3–50,000 points and be at most 2 MB. If a native-resolution export exceeds this, choose an appropriate logarithmic export resolution in REW and follow its smoothing guidance. Do not combine multiple measurements in one text file.
In Open a recording or response, choose the TXT, FRD or response CSV. The page selects Imported response (FRD / REW text). Click Open response data to inspect it.
Use Send to crossover design. The response stays in this tab for up to 30 minutes. In Crossovers, review whether to add a driver or replace an existing branch's response, then click Import into crossover. Existing components and impedance are retained when replacing a response.
Check every driver's measurement position, level, timing reference and actual phase before confirming a common reference. Add measured ZMA impedance where available. Save your project to keep its measurements, components and settings together.
A phase column alone is not proof of measured or shared phase. REW may export 0.0 when phase information is absent. SpeakerFoundry does not verify acoustic timing from text; the crossover workspace resets common-reference confirmation on fresh response import. Room or whole-speaker measurements are not automatically suitable for designing an individual-driver crossover.
Imported response values receive no additional microphone correction or level offset. Phase is unwrapped by equivalent multiples of 360° for interpolation. A two-column response remains magnitude only; the tool does not invent measured phase. A SpeakerFoundry spectrum CSV must not be used as a driver response.
Connect. With amplifier level and input gain low, connect the ECM8000 to Input 1 with a balanced XLR microphone cable. Connect the UMC202HD to the computer by USB. The ECM8000 requires phantom power; enable the interface’s rear +48 V switch after connecting.
Choose the input. In Measurements, allow microphone access and select the UMC202HD input. Check that speaking near the microphone changes the input meter. Keep Direct Monitor off and disable any DAW route returning the microphone to the speaker. Set the system output to the interface or intended playback device; browser audio is not automatically routed through a DAW.
Set a clean level. Increase microphone gain until a moderate test signal is visible without the interface’s CLIP indicator or page clipping warnings. Record the actual browser sample rate and processing warnings. Input selection and channel routing vary by operating system and browser; verify the microphone channel rather than assuming it.
Calibrate or skip. Use a correction file measured for your individual ECM8000 if available. A generic ECM8000 curve does not verify your unit. For optional SPL calibration, play the page’s white noise, take a matching Z / unweighted external reading at the microphone position and enter it. Keep microphone gain and playback settings fixed afterwards. Otherwise use relative level.
Capture twice. Choose a sweep band appropriate for the speaker, start at low amplifier volume and leave the microphone and speaker fixed. Repeat the same capture. Inspect the direct arrival and first reflection before applying a gate; compare raw and gated response before smoothing.
Keep the evidence. Save each raw recording and its FRD/CSV response with setup notes. Send a suitable driver response to the measured crossover workspace. Independent browser sweeps do not establish shared driver phase; reliable acoustic summation needs a verified common timing reference.
Nothing Phone 3a: as of 6 October 2026, we have not found a verified downloadable microphone frequency-response correction file for this model in the checked manufacturer support and measurement-app resources. Use Phone Quick measure for repeatable comparisons with the same phone, position, browser and settings. Its response remains magnitude only.
Level and frequency are separate. The white-noise reference sets a level offset; it does not flatten the microphone across all frequencies. The main measurement page can import an independently verified frequency-error file under advanced microphone correction. Check the exact microphone, orientation, frequency range and sign convention before applying it.
Why an app’s phone compensation is not a universal file
HouseCurve offers internal iPhone/iPad microphone compensation within its app. AudioTool documents personal Android calibration by comparison with a reference microphone, plus files for particular external microphones. Those options do not establish a Nothing Phone 3a correction file suitable for browser capture. Operating-system processing, microphone selection and browser behavior can change the response; a profile needs checking on the actual capture path.
A personal phone profile would need comparison with a reference microphone whose individual correction is known, under controlled and repeatable conditions. An uncalibrated ECM8000 alone cannot establish that absolute correction. Keep the reference recordings and validate the result before using a profile.
Step 2 offers optional external level calibration using the white-noise button and a matching Z / unweighted reference reading. Input gain, microphone and source volume must remain unchanged. Clipping, very quiet input and changes greater than 1 dB between recorded thirds are rejected. The entered offset is external dB SPL minus measured RMS dBFS.
If a microphone frequency-error file is loaded, the calibration capture’s RMS is computed with that correction in the frequency domain, using Parseval’s relation. Its resulting level offset and frequency correction are then applied to browser/external sweep response; changing the frequency file clears the external level calibration. A response-error value is subtracted from the microphone response within the supplied file’s frequency range. Do not use a compensation/gain file with the opposite sign convention.
Load a file for the exact microphone and measurement orientation under Advanced microphone frequency correction & offset. It must contain ascending frequency in Hz and response error in dB. Standard text/CSV/FRD formats are supported.
Take the external reference reading with the same signal and position while capturing. A scalar match does not establish the accuracy of a phone microphone across the frequency range. Do not treat these results as certified sound-level measurements.
Skip calibration for relative dB. A manually entered offset can be useful when independently verified, but does not alone mark a session as externally calibrated.
Imported FRD/REW response text and stereo-reference sweeps retain their existing or relative levels; they are not silently recalibrated by this session’s offset. No microphone correction is applied to an already-calibrated imported response.
Raw/gated response, live spectrum, recording spectrum or stereo transfer
Frequency and level; transfer also includes phase and coherence. Useful for inspection and keeping your measurement record.
FRD
Raw/gated sweep response, qualified stereo transfer or imported response
Frequency Hz, magnitude dB and phase degrees. Transfer exports include only adequately excited reference bins with coherence ≥ 0.80 and require your common-reference/calibrated-level confirmation. Imported response exports preserve phase-column presence and do not imply calibration or common timing.
WAV
Create a test sound
Mono PCM sweep for playback. It is a test signal, not a measured response.
Measure every driver on the intended baffle using the same microphone position, reference voltage, level calibration and timing reference. Maintain a consistent setup when repeating recordings.
Analyse each simultaneous stereo recording, inspect coherence and use Send to crossover design or export its FRD. Name files with the driver model, date, reference voltage and microphone position.
Open Measured-data workspace, select Add driver, import its FRD and a separately measured impedance ZMA where available. Missing ZMA uses the entered resistive approximation and is flagged.
Add series/shunt R, L and C components for each driver. Enter component losses and keep polarity or delay consistent with the measurements.
Confirm a common reference only when it is true, then simulate. Acoustic summation requires phase for every imported driver. The supported frequency range is the overlap of your imported data.
For the Automatic component optimizer, set the target level/slope, frequency band and minimum sampled impedance; lock parts that must stay fixed. Review the candidate before applying it and save your project.
Measure the assembled speaker and compare it with the prediction. Reflections, real component tolerances and driver variation can affect the result.
The optimizer changes component values in your existing network. It does not create an arbitrary topology, prove a global optimum or replace measurement of the finished loudspeaker.
When something does not work
Microphone permission denied or unavailable
Use the HTTPS site, allow microphone access in browser settings and check whether another app holds the input. You can still import an audio recording. Device support varies.
No microphone names in the list
Start capture and grant permission first. The browser may reveal device names only after permission. Stop before choosing another input, then start again.
Microphone stops when changing apps
That is expected: capture/playback stops when the page is backgrounded. Return and start it again.
No data or insufficient audio
Check for silence and correct input/channel. A recording must contain at least one complete 8192-sample window. Check that the browser can decode the selected format.
Transfer analysis asks for stereo
A mono microphone file is insufficient. Record microphone and electrical reference simultaneously in two channels using one clock. Two independent mono recordings cannot establish a dependable common phase reference.
Low coherence or missing frequency bins
Check excitation coverage, reference channel/level, clipping, noise and the repeatability of the recording. A quiet or narrow-band reference may leave insufficient usable data. Re-record with suitable broadband excitation.
FRD export is blocked
Analyse a stereo transfer and confirm the common reference only if valid. At least three coherent frequency bins within 20 Hz–20 kHz must remain. A live or single-channel spectrum exports as CSV only.
Results unexpectedly shift after calibration
Check the response-error sign, frequency units and level offset. Clear correction after changing microphones and analyse again.
Playback goes to the wrong device
Select the intended output in your operating system's sound settings. The browser uses that output; this page does not provide an output-device selector.
File or input settings changed
The previous result is cleared to prevent exporting data with stale settings. Capture or analyse again before export.
Your measurement: microphone → calibration → sweep → response
Microphone & source. Open Measurements. Choose your microphone and the device playing the sound. Use Allow microphone & list devices to reveal inputs; permission is released after listing. A phone normally uses its built-in microphone with an external source. A PC can use a measurement microphone/interface and browser output, or play files through a DAW. Browser audio does not automatically enter a DAW. Use Choose browser output where supported, otherwise system sound settings. Avoid Bluetooth for timing-sensitive work.
Calibrate level, or skip. Play the webpage’s white-noise button through the loudspeaker with low amplifier volume, or download its WAV for your external source. This is band-limited white noise, 500 Hz–2 kHz at nominal −30 dBFS RMS. Choose a band/protection suitable for the driver. Place your reference meter and microphone together. Read the meter using unweighted/Z weighting, enter that reading, confirm steady noise and fixed gain, then Capture noise & calibrate microphone. Noise stops automatically after 30 seconds. An A- or C-weighted broadband reading is not interchangeable with unweighted RMS. If the meter cannot provide Z / unweighted, skip this noise calibration and use relative dB, or enter an independently established offset under advanced settings.
Set up sweep. Set duration, start/end frequencies and digital peak level. Keep the driver and microphone clear of nearby reflecting surfaces, at a fixed distance and orientation. Confirm low volume and a suitable band; an unprotected tweeter needs a suitably high starting frequency. Browser-source recording captures microphone PCM and the digital emitted reference together. Live sweep capture requires at least 6 dB of RMS energy increase over the leading background segment; insufficient signal, clipping and truncated timing are rejected. It does not measure the physical electrical output path.
External player / DAW. Download the exact sweep from this step. Keep its duration, speed, frequency band and EQ unchanged. Start recording on the microphone device, then play that file once from your other source within three seconds of the “play now” message. Capture includes extra time for manual playback. SpeakerFoundry compares the microphone with a generated copy of the known sweep; the external player is not on the microphone’s clock. Clock drift, changed file settings or source DSP can distort the response. This mode is approximate and cannot establish a common phase reference. Verify results against a shared-clock stereo microphone/electrical-reference recording before combining drivers. You can open that stereo WAV from the sweep setup instead.
Raw response first. The Response step opens with the full recovered response, before time gating or smoothing. Room reflections are included. Raw microphone and reference samples stay in memory. Save raw recording & notes to keep a 32-bit float stereo WAV: microphone left, digital or recorded reference right. Reopening this WAV uses relative transfer level and does not automatically restore session SPL calibration.
Apply a gate if needed. Inspect the recorded impulse below the raw result. The largest arrival is selected automatically, but a reflection can be larger than the direct sound. Choose the direct-arrival peak manually when needed. Set the left and right boundaries, ending before the first reflection. A 10% cosine taper softens both edges; the original recording time origin is retained. Apply gate to response, or Show raw response to compare. The gate does not peak-align independent drivers.
Check the usable band. Gated export omits bins below 2 / total gate duration. A 6 ms window has nominal resolution about 167 Hz and export starts around 333 Hz. A short window cannot recover bass resolution. Use a longer reflection-free window or separately measured bass response. Finite sweep bandwidth and windowing can affect endpoints; leave margin around the band you intend to use.
Interpret and export. Review the raw or gated graph, choose magnitude smoothing and graph limits, optionally shift its level, then save CSV/FRD or Send to crossover design. Keep timing unconfirmed until you have verified a common reference. Equal microphone positions or phase columns alone do not prove common timing between separate captures.
Physical capture, device clock stability and SPL accuracy need checking on your microphone/interface. Signal processing and stored-recording workflows are verified with synthetic signals. This is an original SpeakerFoundry workflow for driver development; room decay/RT60 and distortion analysis are not included.
Review, smooth and scale the result
Results offers no smoothing or 1/48, 1/24, 1/12, 1/6 and 1/3-octave magnitude smoothing. Smoothing averages linear power in a logarithmic frequency neighbourhood; original phase samples remain unchanged. Raw response data remains in memory. Prefer light or no smoothing when designing narrow crossover corrections.
Automatic graph limits follow the data; Custom limits lets you enter minimum and maximum dB. These graph limits change only the display. Level shift applies a constant dB adjustment to the plot and exported/transferred data. It is not a substitute for microphone calibration.
Exports are unsmoothed by default. Select Include the selected magnitude smoothing in exported / transferred data if desired. CSV/FRD comments describe processing; gated FRD also includes the gate and timing/level caveats. Reset processing to raw restores an unsmoothed curve, zero level shift and automatic graph limits without re-recording.
Scope, interpretation and your files
Live and single-channel plots are relative per-bin spectra, with an optional entered offset and frequency correction. They are not automatically absolute SPL, power spectral density or isolated driver response. Stereo H1 transfer estimates measurement/reference magnitude and phase; it does not measure impedance.
Recorded sweep deconvolution, impulse-window gating and level-reference calibration are available. Room decay/RT60, distortion analysis and a full room-measurement suite are not included. Choose a recording and calibration method appropriate to your intended result.
All audio analysis and crossover calculations run locally. SpeakerFoundry does not upload or acquire your imported recordings, measurements or designs. Export results and use the project Save controls before leaving. Named device saves stay in that browser and can be lost if browser data is cleared; download a JSON backup to move or keep projects elsewhere. Your own measurements and designs remain yours; share source material only when you have the right to share it.
Measured crossover design and optimizerImport FRD and ZMA driver measurements, build series/shunt RLC branches and optimize unlocked components. Preserve reference settings and save your project.
Passive speaker crossover calculatorsChoose passive speaker crossover order and alignment, calculate L/C values, then add L-pad, notch and baffle-step correction in a complete network.