Midterm: Operation Blind Spot

Overview⌗

The midterm for this course is an adversarial simulation. You will work in two teams: each team builds a fake audio evidence package on Tuesday and analyzes the other team’s package on Thursday.

By the end of Thursday, you will have played both roles that define real forensic audio work — investigator and perpetrator.⌗

What You Are Being Graded On⌗

You receive two separate grades for this midterm:

Tuesday: Building the Fake Case (75 min)⌗

Your team fabricates a realistic-sounding fake audio evidence package. The other team will try to detect your forgeries on Thursday — your job is to make it as convincing as possible.

Step 1 — Record or Assemble Source Audio⌗

You need a base audio recording of at least 2–3 minutes of continuous speech or ambient sound. You have three options:

• Record it now in or near the classroom (hallway, bathroom, stairwell for reverb; treated closet or carpeted space for dry audio)
• Use a public-domain recording from LibriVox or Open Speech Repository (download as WAV)

Save your source material as a 44.1 kHz, 16-bit, mono WAV file before doing anything else.

Step 2 — Apply Required Manipulations⌗

You must apply at least one technique from each category below.

Edit Technique (choose ≥ 1):

Authentication Element (choose ≥ 1):

Aim to conceal your edits. The more convincing your forgery, the better your fabrication grade.

Step 3 — Write the Case Brief⌗

Write a one-page Case Brief that the opposing team will receive on Thursday. The brief must:

• State who allegedly made the recording and who is named in it
• State when the recording was allegedly made (specific date and approximate time)
• State where the recording was allegedly made (indoor/outdoor, general location)
• State what the recording allegedly proves — the legal or factual claim it supports (e.g., an alibi, a threat, a confession, a timestamped presence at a location)

Bias warning: The Case Brief must not tell the analysis team what to listen for, describe the audio contents in detail, or suggest what conclusions they should reach. Give them context without priming. Follow the same principles discussed in Week 3 (LSU — Linear Sequential Unmasking). If your brief would make an unprimed analyst more likely to reach a particular conclusion before doing any technical analysis, revise it.

Good example:

“This 3-minute WAV file was submitted as alibi evidence in a civil dispute. The recording was allegedly made indoors at the subject’s home on the evening of February 24, 2026. The submitting party claims the recording demonstrates continuous presence at that location.”

Bad example (biased — do not do this):

“This recording sounds a little funny around the 1:20 mark. We think it was spliced to remove something. There’s also something weird in the metadata.”

Step 4 — Assemble the Evidence Package⌗

Create a folder called Team_[A or B]_Evidence containing:

• recording.wav — your manipulated audio file
• case_brief.pdf or case_brief.docx — your one-page Case Brief
• chain_of_custody.docx — filled out (template below)

Submit this folder to D2L. Each team member may submit the same fabrication package. Do not share it with the other team yet — they receive it Thursday morning.

Step 5 — Submit Your Answer Key⌗

In a separate document (not shared with the other team), record:

• Every manipulation you applied, with exact timestamps
• How you attempted to conceal each manipulation
• Whether the correct authenticity determination is “authentic,” “not authentic,” or “could be either” (if you used the red herring approach)

Submit the Answer Key directly to D2L.

Thursday: Forensic Analysis (75 min)⌗

You receive the opposing team’s evidence package at the start of class. Your task is to conduct a systematic forensic examination and submit a formal report by the end of class. You may work together as a team during the analysis, but each team member must submit their own individual report.

Your Job⌗

Analyze the package as a neutral forensic examiner — not as an advocate for any position. Follow the methods from Weeks 4–6. Your report must be grounded in technical evidence, not intuition.

Required Analysis Methods⌗

You must demonstrate at least one method from each of the following categories. If a method produces no findings, document that as well — a negative finding is a valid forensic result.

1. Aural Evaluation

• Listen to the full recording once in quiet conditions at a moderate level
• Note preliminary impressions (foreground sounds, background environment, overall quality)
• Avoid looping sections before completing the full listen — this can create false percepts

2. Waveform Analysis (Adobe Audition)

• View the full waveform and look for visible discontinuities, amplitude jumps, or level inconsistencies
• Zoom in on any suspicious areas to the sample level

3. Spectrogram Analysis (Adobe Audition)

• Switch to Spectral Frequency Display (Shift+D)
• Look for: vertical broadband artifacts (edit points), missing reverb tails, broken horizontal lines in the low-frequency range, noise floor changes
• Use FFT Size 4096 or 8192 for better frequency detail on subtle artifacts

4. Edit Detection

• Based on your waveform and spectrogram findings, identify any candidate edit locations
• Document each with a timestamp, the method that revealed it, and your confidence level (high / medium / low)
• Note the type of edit (butt splice, crossfade, reverb mismatch, noise inconsistency) if determinable

5. Metadata Examination (ImHex)

• Open the WAV file in ImHex
• Inspect the file header: sample rate, bit depth, format tag, any embedded timestamp or device info
• Note anything that is inconsistent with the stated recording conditions in the Case Brief

6. ENF Analysis (openENF)

Run the ENF analysis or write a reasoned explanation:

If openENF is working:

• Run: enf recording.wav
• Report the match score, estimated timestamp, and grid identified (or “no match found”)
• Compare to the scenario’s claimed date and location

If openENF is not working on your machine:

• Write a paragraph explaining: (a) what ENF analysis would involve for this file, (b) what result you would expect based on the scenario’s claims, and (c) what recording conditions would need to be true for ENF to be detectable

Forensic Report Format⌗

Submit a single Word document (.docx) structured as follows:

FORENSIC AUDIO EXAMINATION REPORT

Case Reference: [receiving team name] — Midterm Examination
Examiner(s): [your team names].
Date of Examination: February 26, 2026.
Evidence Received: [file name, format, duration, chain of custody notes].

1. Executive Summary

One paragraph. State your authenticity determination (authentic / not authentic / inconclusive), your confidence level, and the primary technical basis for your conclusion. Write this section last — after completing all analysis.

2. Evidence Received

• File name and format
• Duration
• Sample rate and bit depth (from ImHex header examination)
• Chain of custody: how the file was received, any handling notes

3. Methods Used

Brief list of tools and techniques applied (e.g., “Adobe Audition waveform/spectrogram analysis, ImHex metadata examination, openENF analysis”).

4. Findings

Organized by analysis type. Include screenshots annotated with timestamps or arrows where relevant.

4a. Aural Evaluation What you heard. Note any preliminary indicators, foreground/background sounds, and quality assessments.

4b. Waveform Analysis Document any findings. Screenshot waveform at relevant points.

4c. Spectrogram Analysis Document any findings. Screenshot spectrogram at relevant points.

4d. Edit Detection List each candidate edit location with timestamp, detection method, type, and confidence.

4e. Metadata Examination Report header values and note any inconsistencies with the Case Brief.

4f. ENF Analysis Report openENF results, or write the conceptual analysis paragraph.

5. Opinion

A supported statement of your authenticity determination. Cite the specific findings from Section 4 that support your conclusion. Do not introduce new evidence here.

Appendix: Screenshots

Number all screenshots and reference them from the Findings section (e.g., “Fig. 1”).

Bias reminder: Do not write conclusions before completing the technical analysis. Do not let the scenario’s claims substitute for evidence. Mark any content you cannot understand as [Inaudible] rather than guessing. You are educating the court, not advocating for a position.

Post-Class Debrief⌗

After reports are submitted, I will reveal each team’s Answer Key. Each fabrication team will walk the class through what they embedded and how they tried to conceal it. Then the analysis teams share what they found versus what was actually there.

Submission Checklist⌗

Tuesday (before leaving class) — one submission per team member (identical submissions OK):

• Team_[A or B]_Evidence folder submitted to D2L
  • recording.wav
  • case_brief.pdf or case_brief.docx
  • chain_of_custody.docx
• Answer Key submitted to D2L separately (not in the evidence folder)

Thursday (before leaving class) — individual submission required:

• Forensic report .docx submitted to D2L (each team member submits their own)

Chain of Custody Template⌗

Copy and fill out for your evidence package:

CHAIN OF CUSTODY FORM

Grading Rubric⌗

Fabrication (30 pts)⌗

Forensic Report (70 pts)⌗

Resources⌗

• Adobe Audition User Guide
• Week 3 slides — Ethics and false convictions (bias avoidance, LSU protocol)
• Week 5 slides — Splices and edit detection
• lab.md
• Week 6 slides — ENF analysis and metadata consistency
• Week 6 lab — ENF and metadata