We detected you speak 🇺🇸 English (US). Use MyGhostHunt in English (US)?

MGH Spirit Box

QUICK EVP SESSIONOnline mode

By Martin Ali

Real-time EVP sweeper · measured microphone input · local evidence capture

MGH Spirit Box
Choose your investigationSelect how you want to use the Spirit Box
Connected · online features and automatic sync available
Field readiness · Storage, sync & device checks
Offline Readiness
Prepare app assets before going offline
NOT PREPARED
No offline information waiting to sync
App assets
Preparation needed
Waiting to sync
0 queued
Local storage
10.0 GB free
Confirmed synced
0 items

Vocabulary, phonetic/noise tools, microphone capture, supported device sensors and local evidence logging remain available. Live Radio requires internet.

Device check
Check your microphone, storage and connection. Optional sensors are not required.
READY WITH WARNINGS
Microphone
Permission will be requested when needed
Local storage
10.0 GB free
Connection mode
Online · reconnect protection active
Evidence recovery
Queue clear
Device sensors
APIs available: motion, orientation · a live reading and permission are still required before MGH records sensor evidence
Live Radio
Network available

Field recording tip: Keep MGH in the foreground and the screen awake while capturing evidence. Some phones and browsers suspend audio or sensors when the app is backgrounded or the screen locks. Test a short recording and playback before an important investigation; do not assume background capture is supported.

Investigation can continue; review warnings first.
Guest mode — the Spirit Box works fully, but sessions aren't saved.
Pre-Session Calibration
Microphone
Pending permission
Accelerometer
Ready
Orientation
Ready
Magnetometer
Unavailable
Voice Synthesis
Ready
Echo ready to scan.

• Magnetometer not available on this device — magnetic baseline not tracked.

Calibration never claims a sensor exists if the device doesn't provide it.

MGH SPIRIT BOXLIVE FIELD INSTRUMENT
SYSTEM READY
INPUT STANDBY
EVIDENCE ARMED
MGH measured-signal monitor○ INPUT IDLE
STANDBYINPUT 0
REAL MIC WAVE
Audio level display
SCANNING
Standby
GUARD
Standby
Signal
0
Timer
00:00
Input
Ready to start
READY
VOCAB: Loading… Words
WORDS: 0
HITS: 0
Session & recording
AUDIO CAPTURE00:00 / 00:00

Start a session to record. Playback controls unlock after saving.

Press START to begin
Quick Session

Ready without the clutter

Word mode · clean Vocabulary engine · automatic real-microphone calibration · motion contamination guard · evidence logging. Random chance never creates an automatic hit.

Recommended
InputAuto calibrates
HitsMeasured events
EvidenceTimestamped
IntegrityHash protected

Saved evidence preserves its original timestamp and records a SHA-256 integrity hash when supported. Contamination events stay labelled separately from clean Potential Hits.

RAW MIC · SEPARATE CAPTURE
EVIDENCE · SYNC READY
LOCAL RECORDINGS & PRECISE REVIEW · 0

Original microphone recordings, generated audio masters and listening copies must remain distinguishable. Speaker output can still enter the microphone acoustically: use headphones or mute generated output during evidence capture.

No local recordings yet. Start a session with microphone permission; recording starts automatically unless voice-activated capture is selected.

REPEATABLE AUDIO BENCHMARK

Run identical, independently labelled recordings of silence, short knocks, human speech, device handling and speaker feedback. Measures technical event detection, not paranormal accuracy. Stop the live session first.

Labels: an array of {kind:"transient"|"speech"|"handling"|"silence",startMs:0,endMs:100}. Label known ordinary sounds; never use the app’s own detections as ground truth.

Recording playbackNo recording
48 kHz target · 256 kbpsApplies next session
Recent output audit
No output yet — press START.
Sensor Monitor
Audio Level
0
Motion
Ready
Heading
Ready
Magnetometer
Not available

Sensors reflect real device readings only. No simulated values are shown. Sudden movement or magnetic interference is logged as possible contamination, not paranormal evidence.

Custom Investigator Vocabulary

Your custom words are blended into the scan (25% bias) and saved to EchoVocabulary.

Community feedback

Spirit Box Experiences

Share how the MGH Spirit Box worked for you and add investigation photographs.

Sign in to leave a review
Profile name and date added automatically
Up to 4 images · 8 MB each
Pensacola Paranormal Research Group
Sep 4, 2026

Love how it work and the different options there are. Thus will be very useful in the field

MGH Spirit Box is an experimental ITC-style tool. Outputs may have ordinary explanations and are not proof of paranormal communication. Don't trespass or enter unsafe locations. Not a substitute for professional or emergency advice.

Spirit Box Help

70 topics · Tap to read or jump to a control.

Home / Sign InTop-left control
What it is
Navigation and account access. Your session identity and workspace become easier to organise; setup does not itself start recording.
How to use it
Home returns signed-in users to MGH. Guests see Sign In. Guest sessions can run, but they are not saved to the account.
Why it works
Signing in links saved sessions to your account; Home returns to the main MGH workspace.
When to use it
Before saving account evidence or returning to Home.
Best practices
Confirm the correct account and check queued evidence before leaving.
Bluetooth AudioTop header / mobile AUDIO button
What it is
Audio output, voice or device-routing controls. The phone controls pairing and audio routing; microphone availability depends on the operating system.
How to use it
Opens MGH Bluetooth Audio. Android uses the phone's real Bluetooth settings and detects connected audio outputs. Browser builds use the system/browser output chooser when supported. Spirit Box speech follows the phone's active media route.
Why it works
The operating system routes media and supplies supported voices; device availability and volume affect audibility.
When to use it
Before fieldwork and after changing speakers, headphones or voice settings.
Best practices
Pair in system settings where needed, test the route and begin at a comfortable volume.
Evidence iconTop header
What it is
Saved evidence, investigation information and reporting. Saved records show captured information. Older records may have missing metadata; sync requires connectivity and the correct account.
How to use it
Opens saved Spirit Box sessions and evidence. Use it to replay previous work rather than starting a new scan.
Why it works
Stored recordings, timestamps and metadata connect observations to a session; reports reproduce saved information.
When to use it
Before a session for identity details, after capture for review and before sharing.
Best practices
Add attendees, date, time, location, equipment and conditions; check sources, missing fields and sync status.
Help iconTop header
What it is
Searchable instructions and troubleshooting. The full manual gives detailed instructions and troubleshooting.
How to use it
Opens this Spirit Box instruction manual.
Why it works
Topic search connects control names to explanations and the relevant workspace.
When to use it
Whenever a control is unfamiliar or a check fails.
Best practices
Search the exact control name, read its limits, jump to the area and close Help when finished.
Session NameBefore a session
What it is
Saved evidence, investigation information and reporting. Your session identity and workspace become easier to organise; setup does not itself start recording.
How to use it
Optional label for the session. Use a clear case name, room or investigation number.
Why it works
Stored recordings, timestamps and metadata connect observations to a session; reports reproduce saved information.
When to use it
Before a session for identity details, after capture for review and before sharing.
Best practices
Add attendees, date, time, location, equipment and conditions; check sources, missing fields and sync status.
LocationBefore a session
What it is
Saved evidence, investigation information and reporting. Your session identity and workspace become easier to organise; setup does not itself start recording.
How to use it
Optional location label saved with the investigation record.
Why it works
Stored recordings, timestamps and metadata connect observations to a session; reports reproduce saved information.
When to use it
Before a session for identity details, after capture for review and before sharing.
Best practices
Add attendees, date, time, location, equipment and conditions; check sources, missing fields and sync status.
Field ModeMain mobile instrument
What it is
A Spirit Box workspace control. Outputs depend on the selected engine, real input, permissions and supported sensors. A quiet session may have no potential hits.
How to use it
The streamlined investigation layout keeps the evidence display, START/STOP, live MIC/REC/TIMER/EVIDENCE status and essential field actions together. Automatic calibration and raw-evidence protection remain active.
Why it works
This control performs the documented interface action; results depend on device permissions and available session data.
When to use it
When carrying out the corresponding step in setup, investigation or review.
Best practices
Follow the steps below, verify the result and preserve original evidence.
TUNEField Mode rail
What it is
A Spirit Box workspace control. Outputs depend on the selected engine, real input, permissions and supported sensors. A quiet session may have no potential hits.
How to use it
Opens/prepares the advanced professional controls while the session is stopped. It does not delete evidence or start a session.
Why it works
This control performs the documented interface action; results depend on device permissions and available session data.
When to use it
When carrying out the corresponding step in setup, investigation or review.
Best practices
Follow the steps below, verify the result and preserve original evidence.
Use functions separatelyField instrument / navigation
What it is
A choice of investigation workspace. Your session identity and workspace become easier to organise; setup does not itself start recording.
How to use it
Launch Sensor Link, EVP Lab, Evidence Vault, MGH Connect, Field Kit, Settings or the User Manual independently. You never have to run every tool at once.
Why it works
The workspace exposes recommended or advanced controls without requiring every tool to run.
When to use it
Before choosing a simple session, a controlled investigation or an individual tool.
Best practices
Choose the simplest workspace for your task, test input and keep existing evidence.
Field KitStart panel / header briefcase
What it is
Field documentation and supporting case evidence. You can add case information and deliberately capture or choose evidence images.
How to use it
Opens the optional case workspace and Evidence Camera without changing or stopping the current Spirit Box session.
Why it works
Original images and factual notes preserve scene context alongside recordings.
When to use it
During site documentation and when organising a case.
Best practices
Keep originals, record who/where/when and separate observations from interpretations.
Save CaseField Kit
What it is
Saved evidence, investigation information and reporting. You can add case information and deliberately capture or choose evidence images.
How to use it
Stores the case/session name, location, investigation team, witness notes and case notes on this device. During an active session MGH also adds a timestamped Field Kit marker to the investigation timeline.
Why it works
Stored recordings, timestamps and metadata connect observations to a session; reports reproduce saved information.
When to use it
Before a session for identity details, after capture for review and before sharing.
Best practices
Add attendees, date, time, location, equipment and conditions; check sources, missing fields and sync status.
Take PhotoField Kit · Evidence Camera
What it is
Field documentation and supporting case evidence. You can add case information and deliberately capture or choose evidence images.
How to use it
Opens the device camera only when you tap it. MGH does not start the camera automatically.
Why it works
Original images and factual notes preserve scene context alongside recordings.
When to use it
During site documentation and when organising a case.
Best practices
Keep originals, record who/where/when and separate observations from interpretations.
Choose PhotoField Kit · Evidence Camera
What it is
Field documentation and supporting case evidence. You can add case information and deliberately capture or choose evidence images.
How to use it
Selects an existing original image from the device for evidence review.
Why it works
Original images and factual notes preserve scene context alongside recordings.
When to use it
During site documentation and when organising a case.
Best practices
Keep originals, record who/where/when and separate observations from interpretations.
Save Original to Evidence VaultField Kit · Evidence Camera
What it is
Saved evidence, investigation information and reporting. You can add case information and deliberately capture or choose evidence images.
How to use it
Uploads the selected original image to your private Evidence Vault and stores its SHA-256 integrity hash when the device supports hashing. The original image is not edited by this action.
Why it works
Stored recordings, timestamps and metadata connect observations to a session; reports reproduce saved information.
When to use it
Before a session for identity details, after capture for review and before sharing.
Best practices
Add attendees, date, time, location, equipment and conditions; check sources, missing fields and sync status.
Full GuideHelp / User Manual
What it is
Searchable instructions and troubleshooting. The full manual gives detailed instructions and troubleshooting.
How to use it
Opens this complete manual from the Spirit Box screen.
Why it works
Topic search connects control names to explanations and the relevant workspace.
When to use it
Whenever a control is unfamiliar or a check fails.
Best practices
Search the exact control name, read its limits, jump to the area and close Help when finished.
Re-checkPre-Session Calibration
What it is
Input calibration and event-detection settings. Outputs depend on the selected engine, real input, permissions and supported sensors. A quiet session may have no potential hits.
How to use it
Runs the capability check again after you connect hardware, change permissions or move to another device.
Why it works
Calibration measures ordinary input so thresholds can distinguish changes. Noise and movement can still trigger events.
When to use it
Before recording, after moving the device or changing the room.
Best practices
Keep the device still and remain quiet; recalibrate after setup changes and compare control periods.
Operating Mode buttonsWord / Phoneme / Phrase / Hybrid / Question / Sensor Link
What it is
A timestamped question and listening workflow. Outputs depend on the selected engine, real input, permissions and supported sensors. A quiet session may have no potential hits.
How to use it
Selects how output is produced. Question Mode opens the Ask Question overlay. MGH Sensor Link is event-driven and waits for a clean supported sensor anomaly rather than cycling words.
Why it works
The tool records question context so later events can be compared with timing; it cannot prove an answer.
When to use it
When documenting a question during an investigation.
Best practices
Ask one neutral question, leave a quiet gap and retain both responses and silence.
Blind Investigation ModeInvestigation Protocols
What it is
A comparison procedure for interpreting session results. Outputs depend on the selected engine, real input, permissions and supported sensors. A quiet session may have no potential hits.
How to use it
Hides generated outputs while observations are written. Lock observations before revealing the timestamped record.
Why it works
Reducing cues or comparing baseline periods helps reveal expectation and ordinary patterns.
When to use it
When testing whether apparent responses repeat under comparable conditions.
Best practices
Keep duration and settings comparable; record the method before reviewing and include unsuccessful trials.
Control SessionInvestigation Protocols
What it is
A comparison procedure for interpreting session results. Outputs depend on the selected engine, real input, permissions and supported sensors. A quiet session may have no potential hits.
How to use it
Runs Investigation and neutral Control periods with matching settings. While active, key controls are locked so the comparison stays fair.
Why it works
Reducing cues or comparing baseline periods helps reveal expectation and ordinary patterns.
When to use it
When testing whether apparent responses repeat under comparable conditions.
Best practices
Keep duration and settings comparable; record the method before reviewing and include unsuccessful trials.
Engine · CustomiseEngine panel
What it is
The selected audio/output mode or vocabulary control. Outputs depend on the selected engine, real input, permissions and supported sensors. A quiet session may have no potential hits.
How to use it
Expands the engine chooser, specialist options and professional presets.
Why it works
The selected engine assembles or plays its configured audio material. Recognisable words can occur through ordinary selection and interpretation.
When to use it
When choosing the type of output to test or comparing controlled sessions.
Best practices
Change one setting at a time, document it and compare results with a control. Generated words are not independent proof of communication.
Vocabulary EngineEngine option
What it is
The selected audio/output mode or vocabulary control. Outputs depend on the selected engine, real input, permissions and supported sensors. A quiet session may have no potential hits.
How to use it
Uses the complete 452k+ generated MGH vocabulary.
Why it works
The selected engine assembles or plays its configured audio material. Recognisable words can occur through ordinary selection and interpretation.
When to use it
When choosing the type of output to test or comparing controlled sessions.
Best practices
Change one setting at a time, document it and compare results with a control. Generated words are not independent proof of communication.
Phonetic EngineEngine option
What it is
The selected audio/output mode or vocabulary control. Outputs depend on the selected engine, real input, permissions and supported sensors. A quiet session may have no potential hits.
How to use it
Uses wordless phoneme, syllable, breath and non-lexical fragment banks.
Why it works
The selected engine assembles or plays its configured audio material. Recognisable words can occur through ordinary selection and interpretation.
When to use it
When choosing the type of output to test or comparing controlled sessions.
Best practices
Change one setting at a time, document it and compare results with a control. Generated words are not independent proof of communication.
Noise EngineEngine option
What it is
The selected audio/output mode or vocabulary control. Outputs depend on the selected engine, real input, permissions and supported sensors. A quiet session may have no potential hits.
How to use it
Generates a local noise bed. It does not create words; microphone detection still runs separately.
Why it works
The selected engine assembles or plays its configured audio material. Recognisable words can occur through ordinary selection and interpretation.
When to use it
When choosing the type of output to test or comparing controlled sessions.
Best practices
Change one setting at a time, document it and compare results with a control. Generated words are not independent proof of communication.
Live RadioEngine option
What it is
Internet-stream audio used by the selected engine. Outputs depend on the selected engine, real input, permissions and supported sensors. A quiet session may have no potential hits.
How to use it
Uses a user-supplied live internet stream. It is not a physical AM/FM receiver and requires internet.
Why it works
A network stream supplies broadcast audio; speech may come from ordinary radio content.
When to use it
When connected and deliberately testing a live-stream engine.
Best practices
Check the stream and volume; document the station and distinguish broadcast words from microphone evidence.
MGH Advanced Phonetic SweepEngine option
What it is
The selected audio/output mode or vocabulary control. Outputs depend on the selected engine, real input, permissions and supported sensors. A quiet session may have no potential hits.
How to use it
High-performance locally generated micro-phonetic engine with 25–500ms sweep timing, 5–15ms crossfades, reverse fragments, optional dual-channel overlap, high/low-pass filters, echo feedback, binaural reverb, environmental entropy seeding and separate lossless sweep-master telemetry.
Why it works
The selected engine assembles or plays its configured audio material. Recognisable words can occur through ordinary selection and interpretation.
When to use it
When choosing the type of output to test or comparing controlled sessions.
Best practices
Change one setting at a time, document it and compare results with a control. Generated words are not independent proof of communication.
Phonetic BanksPhonetic engine
What it is
The selected audio/output mode or vocabulary control. Outputs depend on the selected engine, real input, permissions and supported sensors. A quiet session may have no potential hits.
How to use it
Choose one or more fragment banks. If none are selected, all banks are used.
Why it works
The selected engine assembles or plays its configured audio material. Recognisable words can occur through ordinary selection and interpretation.
When to use it
When choosing the type of output to test or comparing controlled sessions.
Best practices
Change one setting at a time, document it and compare results with a control. Generated words are not independent proof of communication.
Noise TypeNoise engine
What it is
The selected audio/output mode or vocabulary control. Outputs depend on the selected engine, real input, permissions and supported sensors. A quiet session may have no potential hits.
How to use it
Selects the locally generated noise character such as white, pink or brown noise.
Why it works
The selected engine assembles or plays its configured audio material. Recognisable words can occur through ordinary selection and interpretation.
When to use it
When choosing the type of output to test or comparing controlled sessions.
Best practices
Change one setting at a time, document it and compare results with a control. Generated words are not independent proof of communication.
Professional PresetsEngine panel
What it is
The selected audio/output mode or vocabulary control. Outputs depend on the selected engine, real input, permissions and supported sensors. A quiet session may have no potential hits.
How to use it
Applies a prepared combination such as Clean Investigation, Traditional Sweep, Wordless Phonetic, EVP Listening, Quiet Location or Noisy Location.
Why it works
The selected engine assembles or plays its configured audio material. Recognisable words can occur through ordinary selection and interpretation.
When to use it
When choosing the type of output to test or comparing controlled sessions.
Best practices
Change one setting at a time, document it and compare results with a control. Generated words are not independent proof of communication.
RefreshReal microphone input
What it is
Real microphone input and device selection. Outputs depend on the selected engine, real input, permissions and supported sensors. A quiet session may have no potential hits.
How to use it
Refreshes the list of microphones exposed by the phone or browser.
Why it works
The browser captures the permitted audio device. Room noise, handling and routing affect the captured signal.
When to use it
Before starting a session or whenever the microphone changes.
Best practices
Grant permission, refresh/select the device, test it and keep generated playback separate from original microphone evidence.
Microphone selectorReal microphone input
What it is
Real microphone input and device selection. Outputs depend on the selected engine, real input, permissions and supported sensors. A quiet session may have no potential hits.
How to use it
Chooses System default or an exposed USB, Bluetooth or wired microphone. It is locked while a session is running.
Why it works
The browser captures the permitted audio device. Room noise, handling and routing affect the captured signal.
When to use it
Before starting a session or whenever the microphone changes.
Best practices
Grant permission, refresh/select the device, test it and keep generated playback separate from original microphone evidence.
TEST VOICEVoice diagnostics
What it is
Audio output, voice or device-routing controls. A voice test should be audible through the active media output. Installed voices and routing vary by device.
How to use it
Plays a spoken test so you can confirm device volume and text-to-speech before investigating.
Why it works
The operating system routes media and supplies supported voices; device availability and volume affect audibility.
When to use it
Before fieldwork and after changing speakers, headphones or voice settings.
Best practices
Pair in system settings where needed, test the route and begin at a comfortable volume.
Voice DiagnosticsVoice diagnostics
What it is
Audio output, voice or device-routing controls. A voice test should be audible through the active media output. Installed voices and routing vary by device.
How to use it
Opens the detailed speech test page when the normal voice test is not working.
Why it works
The operating system routes media and supplies supported voices; device availability and volume affect audibility.
When to use it
Before fieldwork and after changing speakers, headphones or voice settings.
Best practices
Pair in system settings where needed, test the route and begin at a comfortable volume.
Speed sliderMain Spirit Box
What it is
The scanning-cycle timing control. Outputs depend on the selected engine, real input, permissions and supported sensors. A quiet session may have no potential hits.
How to use it
Controls how quickly scanning cycles. Left is slower; right is faster.
Why it works
The timing setting changes how often output cycles; faster output does not imply stronger evidence.
When to use it
Before comparing slow and fast scans.
Best practices
Change timing alone, log the setting and compare equal-length control periods.
START SESSIONDirectly below the main evidence display
What it is
Session capture and recording controls. Saved records show captured information. Older records may have missing metadata; sync requires connectivity and the correct account.
How to use it
Begins scanning, requests microphone access if needed and starts real-input calibration. On mobile this is deliberately attached to the preview/readout so the primary action is always immediately visible.
Why it works
Recording stores captured audio/events for later review; playback is available only when a recording exists.
When to use it
At the beginning, during breaks and at the end of an investigation.
Best practices
Test input first, note participants and surroundings, stop cleanly and check the saved recording.
STOP SESSIONDirectly below the main evidence display while active
What it is
Session capture and recording controls. Saved records show captured information. Older records may have missing metadata; sync requires connectivity and the correct account.
How to use it
Replaces START in the same primary position, ends the active session and finalises the session record.
Why it works
Recording stores captured audio/events for later review; playback is available only when a recording exists.
When to use it
At the beginning, during breaks and at the end of an investigation.
Best practices
Test input first, note participants and surroundings, stop cleanly and check the saved recording.
Pause / ResumeDuring a session
What it is
Session capture and recording controls. Starting requests required permissions and calibrates input; pause/resume preserves the session and stop finalises it.
How to use it
Temporarily pauses or resumes scanning without ending the session.
Why it works
Recording stores captured audio/events for later review; playback is available only when a recording exists.
When to use it
At the beginning, during breaks and at the end of an investigation.
Best practices
Test input first, note participants and surroundings, stop cleanly and check the saved recording.
Record / Stop RecRecording playback
What it is
Session capture and recording controls. Only available recordings can be replayed. A manual mark flags something for review; it does not confirm its cause.
How to use it
Starts or stops a separate session audio recording. It is available only while the session is running.
Why it works
Recording stores captured audio/events for later review; playback is available only when a recording exists.
When to use it
At the beginning, during breaks and at the end of an investigation.
Best practices
Test input first, note participants and surroundings, stop cleanly and check the saved recording.
-10sRecording playback
What it is
Recorded-session playback and review. Only available recordings can be replayed. A manual mark flags something for review; it does not confirm its cause.
How to use it
Rewinds the saved recording by ten seconds.
Why it works
Playback revisits captured sound and timestamps; repeated listening can increase expectation bias.
When to use it
After recording, to inspect an event and its surrounding context.
Best practices
Use normal speed first, safe volume and independent listener notes.
Play / PauseRecording playback
What it is
Recorded-session playback and review. Only available recordings can be replayed. A manual mark flags something for review; it does not confirm its cause.
How to use it
Plays or pauses the current recording after one exists.
Why it works
Playback revisits captured sound and timestamps; repeated listening can increase expectation bias.
When to use it
After recording, to inspect an event and its surrounding context.
Best practices
Use normal speed first, safe volume and independent listener notes.
+10sRecording playback
What it is
Recorded-session playback and review. Only available recordings can be replayed. A manual mark flags something for review; it does not confirm its cause.
How to use it
Moves the recording forward ten seconds.
Why it works
Playback revisits captured sound and timestamps; repeated listening can increase expectation bias.
When to use it
After recording, to inspect an event and its surrounding context.
Best practices
Use normal speed first, safe volume and independent listener notes.
EVP Playback SpeedEVP Lab · Library
What it is
Recorded-session playback and review. Audio capture and analysis show recorded sound. Processing changes a listening copy, not the original or its meaning.
How to use it
Reviews saved audio at 0.5×, 0.75×, 1×, 1.25×, 1.5× or 2×. Pitch is preserved where the device/browser supports it. This changes listening playback only and never changes the original recording.
Why it works
Playback revisits captured sound and timestamps; repeated listening can increase expectation bias.
When to use it
After recording, to inspect an event and its surrounding context.
Best practices
Use normal speed first, safe volume and independent listener notes.
Waveform / SpectrumEVP Lab · Library
What it is
Recorded-audio analysis or copy-processing tools. Audio capture and analysis show recorded sound. Processing changes a listening copy, not the original or its meaning.
How to use it
Opens the measured full-duration waveform and live spectrogram for a saved EVP. These show audio amplitude and frequency content only; they do not identify paranormal voices.
Why it works
Waveforms show amplitude and spectra show frequency content; processing changes a listening copy and preserves the original.
When to use it
After saving original audio, when inspecting a particular segment.
Best practices
Listen to the original first, retain it and document all processing.
Process CopyEVP Lab · Library
What it is
Recorded-audio analysis or copy-processing tools. Audio capture and analysis show recorded sound. Processing changes a listening copy, not the original or its meaning.
How to use it
Opens the non-destructive EVP editor. It can trim a listening copy, adjust gain, apply high-pass/low-pass filters, normalize peak level or reverse the copy. The raw original is never overwritten.
Why it works
Waveforms show amplitude and spectra show frequency content; processing changes a listening copy and preserves the original.
When to use it
After saving original audio, when inspecting a particular segment.
Best practices
Listen to the original first, retain it and document all processing.
Trim Start / EndEVP Lab · Process Copy
What it is
Recorded-audio analysis or copy-processing tools. Audio capture and analysis show recorded sound. Processing changes a listening copy, not the original or its meaning.
How to use it
Chooses the exact part of the original recording to include in the new derived WAV. Enter times in seconds. The source recording remains unchanged.
Why it works
Waveforms show amplitude and spectra show frequency content; processing changes a listening copy and preserves the original.
When to use it
After saving original audio, when inspecting a particular segment.
Best practices
Listen to the original first, retain it and document all processing.
GainEVP Lab · Process Copy
What it is
Recorded-audio analysis or copy-processing tools. Audio capture and analysis show recorded sound. Processing changes a listening copy, not the original or its meaning.
How to use it
Raises or lowers the derived copy from -24 dB to +24 dB using real Web Audio processing. It can make quiet sound easier to hear but can also exaggerate ordinary noise.
Why it works
Waveforms show amplitude and spectra show frequency content; processing changes a listening copy and preserves the original.
When to use it
After saving original audio, when inspecting a particular segment.
Best practices
Listen to the original first, retain it and document all processing.
High-pass / Low-passEVP Lab · Process Copy
What it is
Recorded-audio analysis or copy-processing tools. Audio capture and analysis show recorded sound. Processing changes a listening copy, not the original or its meaning.
How to use it
Applies real frequency filters to the derived copy. High-pass reduces frequencies below the selected value; low-pass reduces frequencies above the selected value.
Why it works
Waveforms show amplitude and spectra show frequency content; processing changes a listening copy and preserves the original.
When to use it
After saving original audio, when inspecting a particular segment.
Best practices
Listen to the original first, retain it and document all processing.
Normalize PeakEVP Lab · Process Copy
What it is
Recorded-audio analysis or copy-processing tools. Audio capture and analysis show recorded sound. Processing changes a listening copy, not the original or its meaning.
How to use it
Raises the derived copy toward a safe peak level without overwriting the original. Normalization changes listening level, not evidential meaning.
Why it works
Waveforms show amplitude and spectra show frequency content; processing changes a listening copy and preserves the original.
When to use it
After saving original audio, when inspecting a particular segment.
Best practices
Listen to the original first, retain it and document all processing.
Reverse AudioEVP Lab · Process Copy
What it is
Recorded-audio analysis or copy-processing tools. Audio capture and analysis show recorded sound. Processing changes a listening copy, not the original or its meaning.
How to use it
Creates the derived copy with the selected audio segment reversed. This is an experimental listening aid and must not be interpreted as proof.
Why it works
Waveforms show amplitude and spectra show frequency content; processing changes a listening copy and preserves the original.
When to use it
After saving original audio, when inspecting a particular segment.
Best practices
Listen to the original first, retain it and document all processing.
Create Derived WAVEVP Lab · Process Copy
What it is
Recorded-audio analysis or copy-processing tools. Audio capture and analysis show recorded sound. Processing changes a listening copy, not the original or its meaning.
How to use it
Renders the selected processing into a new WAV, stores the processing settings and links it back to the original evidence. Where supported, the new derived file receives its own SHA-256 integrity hash.
Why it works
Waveforms show amplitude and spectra show frequency content; processing changes a listening copy and preserves the original.
When to use it
After saving original audio, when inspecting a particular segment.
Best practices
Listen to the original first, retain it and document all processing.
Mute / UnmuteDuring a session
What it is
Spoken-output volume control. Starting requests required permissions and calibrates input; pause/resume preserves the session and stop finalises it.
How to use it
Mutes or restores Spirit Box spoken output. It does not switch off the real microphone measurement.
Why it works
Muting suppresses generated speech while real microphone measurement remains separate.
When to use it
When listening to the room or reducing generated-audio interference.
Best practices
Verify the microphone remains active and note muted periods.
AskDuring a session
What it is
A timestamped question and listening workflow. Starting requests required permissions and calibrates input; pause/resume preserves the session and stop finalises it.
How to use it
Opens the question recorder/timestamp tool. Ask clearly, then leave a listening gap.
Why it works
The tool records question context so later events can be compared with timing; it cannot prove an answer.
When to use it
When documenting a question during an investigation.
Best practices
Ask one neutral question, leave a quiet gap and retain both responses and silence.
MarkDuring a session
What it is
Potential-event markers and review classifications. Starting requests required permissions and calibrates input; pause/resume preserves the session and stop finalises it.
How to use it
Manually flags the current output as worth reviewing. A manual mark is investigator input, not an automatic hit.
Why it works
Markers preserve a timestamp/source or detected correlation. They do not verify a paranormal cause.
When to use it
When something needs later review or has a known ordinary explanation.
Best practices
Record source and surrounding context; retain contamination flags and avoid upgrading a mark without evidence.
ModeAdvanced control
What it is
A Spirit Box workspace control. Outputs depend on the selected engine, real input, permissions and supported sensors. A quiet session may have no potential hits.
How to use it
Cycles through operating modes without using the large mode grid.
Why it works
This control performs the documented interface action; results depend on device permissions and available session data.
When to use it
When carrying out the corresponding step in setup, investigation or review.
Best practices
Follow the steps below, verify the result and preserve original evidence.
SettingsMain controls
What it is
An interface or session preference. Settings affect interface and audio behaviour; some controls are locked during controlled investigations.
How to use it
Opens interface, audio, notification and instruction settings.
Why it works
The selected option changes its documented display, audio or notification behaviour; it does not increase evidence validity.
When to use it
During setup or when adjusting the interface to your investigation.
Best practices
Change one preference at a time and verify its effect; respect controls locked during an active session.
EvidenceMain controls
What it is
Saved evidence, investigation information and reporting. Saved records show captured information. Older records may have missing metadata; sync requires connectivity and the correct account.
How to use it
Opens the saved-session library.
Why it works
Stored recordings, timestamps and metadata connect observations to a session; reports reproduce saved information.
When to use it
Before a session for identity details, after capture for review and before sharing.
Best practices
Add attendees, date, time, location, equipment and conditions; check sources, missing fields and sync status.
SaveMain controls
What it is
Saved evidence, investigation information and reporting. Saved records show captured information. Older records may have missing metadata; sync requires connectivity and the correct account.
How to use it
Saves the current session record. Signed-in users can keep it with their evidence.
Why it works
Stored recordings, timestamps and metadata connect observations to a session; reports reproduce saved information.
When to use it
Before a session for identity details, after capture for review and before sharing.
Best practices
Add attendees, date, time, location, equipment and conditions; check sources, missing fields and sync status.
ReportAfter a session
What it is
Saved evidence, investigation information and reporting. Saved records show captured information. Older records may have missing metadata; sync requires connectivity and the correct account.
How to use it
Creates the investigation report from a finished session when recorded output exists.
Why it works
Stored recordings, timestamps and metadata connect observations to a session; reports reproduce saved information.
When to use it
Before a session for identity details, after capture for review and before sharing.
Best practices
Add attendees, date, time, location, equipment and conditions; check sources, missing fields and sync status.
MARKMain controls
What it is
Potential-event markers and review classifications. Starting requests required permissions and calibrates input; pause/resume preserves the session and stop finalises it.
How to use it
Explicitly marks the current generated output for investigator review. The instrument background itself does not create marks, reducing accidental evidence tagging.
Why it works
Markers preserve a timestamp/source or detected correlation. They do not verify a paranormal cause.
When to use it
When something needs later review or has a known ordinary explanation.
Best practices
Record source and surrounding context; retain contamination flags and avoid upgrading a mark without evidence.
Play event audioPotential Hit card
What it is
Recorded-session playback and review. Only available recordings can be replayed. A manual mark flags something for review; it does not confirm its cause.
How to use it
Replays the original microphone recording around a measured or manually marked event when a recording exists.
Why it works
Playback revisits captured sound and timestamps; repeated listening can increase expectation bias.
When to use it
After recording, to inspect an event and its surrounding context.
Best practices
Use normal speed first, safe volume and independent listener notes.
Save to VaultPotential Hit card
What it is
Saved evidence, investigation information and reporting. Saved records show captured information. Older records may have missing metadata; sync requires connectivity and the correct account.
How to use it
Saves the reviewed event into the Evidence Vault.
Why it works
Stored recordings, timestamps and metadata connect observations to a session; reports reproduce saved information.
When to use it
Before a session for identity details, after capture for review and before sharing.
Best practices
Add attendees, date, time, location, equipment and conditions; check sources, missing fields and sync status.
FlagPotential Hit card
What it is
Potential-event markers and review classifications. Only available recordings can be replayed. A manual mark flags something for review; it does not confirm its cause.
How to use it
Manually flags the event for later attention.
Why it works
Markers preserve a timestamp/source or detected correlation. They do not verify a paranormal cause.
When to use it
When something needs later review or has a known ordinary explanation.
Best practices
Record source and surrounding context; retain contamination flags and avoid upgrading a mark without evidence.
NotePotential Hit card
What it is
Potential-event markers and review classifications. Only available recordings can be replayed. A manual mark flags something for review; it does not confirm its cause.
How to use it
Adds a factual investigator note to the event timeline.
Why it works
Markers preserve a timestamp/source or detected correlation. They do not verify a paranormal cause.
When to use it
When something needs later review or has a known ordinary explanation.
Best practices
Record source and surrounding context; retain contamination flags and avoid upgrading a mark without evidence.
DismissHit / contamination card
What it is
Potential-event markers and review classifications. Only available recordings can be replayed. A manual mark flags something for review; it does not confirm its cause.
How to use it
Closes the card without changing the original underlying session record.
Why it works
Markers preserve a timestamp/source or detected correlation. They do not verify a paranormal cause.
When to use it
When something needs later review or has a known ordinary explanation.
Best practices
Record source and surrounding context; retain contamination flags and avoid upgrading a mark without evidence.
Custom Vocabulary · AddCustom Investigator Vocabulary
What it is
A Spirit Box workspace control. Starting requests required permissions and calibrates input; pause/resume preserves the session and stop finalises it.
How to use it
Adds a name, place or investigation word to your custom list. Custom words are blended into vocabulary scanning with a 25% bias.
Why it works
This control performs the documented interface action; results depend on device permissions and available session data.
When to use it
When carrying out the corresponding step in setup, investigation or review.
Best practices
Follow the steps below, verify the result and preserve original evidence.
Custom word ×Custom Investigator Vocabulary
What it is
The selected audio/output mode or vocabulary control. Starting requests required permissions and calibrates input; pause/resume preserves the session and stop finalises it.
How to use it
Removes that custom word from the investigator vocabulary.
Why it works
The selected engine assembles or plays its configured audio material. Recognisable words can occur through ordinary selection and interpretation.
When to use it
When choosing the type of output to test or comparing controlled sessions.
Best practices
Change one setting at a time, document it and compare results with a control. Generated words are not independent proof of communication.
Online modeConnection controls
What it is
Connection and evidence-sync controls. The chosen mode has a Selected label. Offline items show waiting, syncing, failed or confirmed synced; confirmation appears only after a successful upload.
How to use it
Select Online to enable network features and queue syncing when connected and signed in. If the signal is lost, offline field operation remains active until it returns.
Why it works
Local storage holds queued items; successful account uploads confirm sync. A connection alone does not confirm an upload.
When to use it
Before fieldwork, after signal loss and when reconnecting.
Best practices
Check Selected mode, queued/failed counts and confirmed sync before relying on the account copy.
Offline modeConnection controls
What it is
Connection and evidence-sync controls. The chosen mode has a Selected label. Offline items show waiting, syncing, failed or confirmed synced; confirmation appears only after a successful upload.
How to use it
Select Offline to keep local tools available and queue captured evidence on the device. Live Radio needs internet. Manual Offline stays selected until you choose Online.
Why it works
Local storage holds queued items; successful account uploads confirm sync. A connection alone does not confirm an upload.
When to use it
Before fieldwork, after signal loss and when reconnecting.
Best practices
Check Selected mode, queued/failed counts and confirmed sync before relying on the account copy.
Sync statusOffline readiness and queued evidence
What it is
Connection and evidence-sync controls. The chosen mode has a Selected label. Offline items show waiting, syncing, failed or confirmed synced; confirmation appears only after a successful upload.
How to use it
Waiting means stored locally; syncing means upload is in progress; failed items remain queued; confirmed synced shows the completed count and time.
Why it works
Local storage holds queued items; successful account uploads confirm sync. A connection alone does not confirm an upload.
When to use it
Before fieldwork, after signal loss and when reconnecting.
Best practices
Check Selected mode, queued/failed counts and confirmed sync before relying on the account copy.
Evidence filtersEvidence library
What it is
Saved evidence, investigation information and reporting. Saved records show captured information. Older records may have missing metadata; sync requires connectivity and the correct account.
How to use it
Search session names, people, location, notes, dates or time. Expand Filters to combine output mode, microphone/sensor/manual source and sessions with hits. Clear filters restores the list.
Why it works
Stored recordings, timestamps and metadata connect observations to a session; reports reproduce saved information.
When to use it
Before a session for identity details, after capture for review and before sharing.
Best practices
Add attendees, date, time, location, equipment and conditions; check sources, missing fields and sync status.
Investigation detailsEvidence library
What it is
Saved evidence, investigation information and reporting. Saved records show captured information. Older records may have missing metadata; sync requires connectivity and the correct account.
How to use it
Open a session, expand Edit investigation information and notes, add attendees, date, local time, location, equipment and conditions, then save. Reports and exports include saved details while original capture timestamps stay unchanged.
Why it works
Stored recordings, timestamps and metadata connect observations to a session; reports reproduce saved information.
When to use it
Before a session for identity details, after capture for review and before sharing.
Best practices
Add attendees, date, time, location, equipment and conditions; check sources, missing fields and sync status.
🍪◉

Essential cookies keep the app working. Optional analytics measure time spent in features for admins only. Choose whether to allow optional cookies; you can change this in Settings.