Final Project: Operation Deep Cut

Overview⌗

The final project is a full-semester capstone simulation. You will work in two teams: each team builds an audio evidence package over two class days, then receives and analyzes the opposing team’s package over the following two days. On final exam day, both teams present — walking through what they built, what they found, and how your individual examiners compared.

You play both roles: case builder and forensic analyst.

Grading⌗

Choosing Your Scenario Type⌗

At the start of Week 15, each team chooses one scenario type. The two teams must choose different types.

You will not analyze the same type you built — you receive the opposing team’s case on Week 16 Tuesday.

Week 15: Case Building (Two Days)⌗

Day 1: Research Your Scenario (Before You Build Anything)⌗

Before recording or editing anything, spend the first part of Tuesday researching real cases in your scenario type. Your case brief and choices must be grounded in how these incidents actually unfold — not invented from scratch.

Shooting teams should research real shooting incident investigations, paying attention to:

• How multi-source recordings are collected (smartphones, body cameras, security cameras)
• What acoustic signatures are actually present and how they vary with distance and environment
• How investigators establish shot count, location, and sequence
• Example: the Maher investigation of the Trump rally assassination attempt from Week 13

CVR teams should research real CVR investigations, paying attention to:

• What kinds of audio degradation are present in real CVR recordings
• How transcription disputes arise and what makes them difficult to resolve
• What specific claims (about timing, wording, crew state) are typically contested
• Example: the Southern Airways Flight 932 CVR investigation from Week 14

Your goal is plausibility. A case brief or scenario that couldn’t happen in the real world will be easy to dismiss. Research first, build second.

Research documentation requirement: each team must submit a short research_notes.pdf or research_notes.docx with at least 2 real sources that informed the scenario. These may include academic papers, investigative reports, court documents, NTSB materials, or credible news coverage. For each source, include 1-2 sentences explaining what it contributed to your case design.

Check-in at end of Day 1: each team gives a one-sentence summary of the real case(s) they found most useful and how it shaped their scenario plan.

Shooting Incident Requirements⌗

Shooting teams must include:

• At least two recordings from different positions (e.g., mic_A.wav, mic_B.wav)
  • At least one recording must be authentic (unmanipulated)
  • At least one recording must be fabricated — manipulated using an edit technique and authentication element below
• At least one gunshot-like acoustic event with identifiable components (muzzle blast, shockwave, or reflections)
• A case brief with a specific claim about shooter location, distance, or number of shots
• An answer key submitted to the instructor only — identifying which recordings are authentic and which are fabricated

The analysis team’s job is to identify which recordings have been tampered with, exclude them as tainted evidence, and base their forensic conclusions only on the authentic recordings. This mirrors real casework, where fabricated evidence is inadmissible.

Edit Techniques (choose ≥ 1, apply to fabricated recording only):

Authentication Elements (choose ≥ 1, apply to fabricated recording only):

Source material: You may record impulsive sounds (claps, slams, balloons) in different spaces to simulate muzzle blast and environment interactions. Label your recordings by position (e.g., mic_A.wav, mic_B.wav).

CVR Incident Requirements⌗

CVR recordings are official flight records — not something a party would realistically fabricate. CVR teams build a realistic degraded recording and a contested scenario claim. No fabricated edits required. Include:

• At least 3–5 minutes of cockpit-style dialogue (you may use public domain speech from LibriVox or record your own)
• A plausible CVR-style source structure: one mixed recording is acceptable, but teams are encouraged to simulate multiple sources or channels (for example, pilot headset, copilot headset, and cockpit area mic) if it helps support the scenario
• Degradation sufficient to require enhancement before interpretation — apply noise, clipping, or heavy reverb
• A case brief with a specific claim about what was said at a particular moment, or when a key statement was made
• An answer key submitted to the instructor only — including a full transcript and the correct interpretation of the contested claim

Writing the Case Brief⌗

Follow the same LSU principles from the midterm. The brief must state:

• Who is speaking or named in the recording
• When the recording was made (specific date and approximate time)
• Where the recording was made
• What the recording allegedly proves or establishes

The brief must not tell the analysis team what to listen for, describe the audio contents in detail, or suggest what conclusions to reach.

Assembling the Evidence Package⌗

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

• Your audio file(s) (WAV format)
• case_brief.pdf or case_brief.docx
• research_notes.pdf or research_notes.docx with at least 2 cited sources and a brief note on how each source informed the scenario

Submit to D2L. Answer key goes to the instructor separately — not in the evidence folder.

Week 16: Analysis (Two Days)⌗

You receive the opposing team’s evidence package at the start of Week 16 Tuesday.

Rule: Independent Analysis First⌗

Each team member analyzes the evidence alone, without consulting teammates, until their individual report is submitted. The inter-examiner comparison only happens after all individual reports are in.

Required Analysis Methods (All Types)⌗

Apply all six methods to every recording in the package. If a method produces no findings, document that — a negative finding is a valid forensic result.

1. Aural Evaluation

• Listen to the full recording once in quiet conditions at a moderate level
• Document preliminary impressions before any technical analysis

2. Waveform Analysis (Adobe Audition)

• Look for discontinuities, amplitude jumps, or level inconsistencies
• Zoom in on suspicious areas to the sample level

3. Spectrogram Analysis (Adobe Audition)

• Use Spectral Frequency Display (Shift+D), FFT Size 4096 or 8192
• Look for: vertical broadband artifacts, missing reverb tails, broken horizontal lines, noise floor changes

4. Edit Detection

• Document each candidate edit location with timestamp, detection method, type, and confidence level
• If no edits are found, document that explicitly

5. Metadata Examination (ImHex)

• Inspect file header: sample rate, bit depth, format tag, embedded timestamp or device info
• Note anything inconsistent with the case brief

6. ENF Analysis (openENF)

• Run openENF and report match score, estimated timestamp, and grid
• Or write a conceptual analysis paragraph if openENF is not working
• If no ENF claim is made in the case brief, note whether ENF analysis is applicable and why

Shooting-Specific Analysis⌗

Step 1 — Authenticate each recording before using it

Apply the six required methods to every recording individually. For each recording, reach a determination: authentic or tainted.

• Tainted recordings must be excluded from acoustic analysis. Document why and state that the recording cannot be used as evidence.
• Only authentic recordings may be used for acoustic event analysis and TDOA.

Step 2 — Acoustic and TDOA analysis (authentic recordings only)

• Identify and classify acoustic events (muzzle blast, shockwave, reflections) in each authentic recording
• Measure the arrival time difference of the gunshot event between recordings in Adobe Audition — zoom to the sample level to identify the precise onset in each file, then calculate the time delay between them
• Use the time delay and the speed of sound (adjusted for temperature if known) to estimate the shooter’s distance or position
• Express your conclusion using likelihood ratio framing
• Evaluate the scenario’s specific claim about location, distance, or shot count

Executive Summary for shooting cases: State which recordings you determined to be authentic and which tainted, your basis for each determination, and what the authentic recordings support about the scenario’s claim.

CVR-Specific Analysis⌗

• Determine whether enhancement is necessary for interpretability
• If yes: apply enhancement in Adobe Audition, document your before/after settings, and justify the approach
• Transcribe key passages; mark unintelligible content as [Inaudible]
• Evaluate the scenario’s specific claim about what was said or when

Executive Summary for CVR cases: State your interpretation of the contested claim, your confidence level, and the primary technical basis (enhancement decisions, transcription, acoustic analysis).

Individual Report Format⌗

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

FORENSIC AUDIO EXAMINATION REPORT

Case Reference: [receiving team name] — Final Project
Examiner: [your name]
Date of Examination: [date]
Evidence Received: [file names, format, duration]

1. Executive Summary

One paragraph. See scenario-specific guidance above. Write this last.

2. Evidence Received

File name(s), format, duration, sample rate, bit depth.

3. Methods Used

Brief list of tools and techniques applied.

4. Findings

Organized by analysis type with annotated screenshots.

• 4a. Aural Evaluation
• 4b. Waveform Analysis
• 4c. Spectrogram Analysis
• 4d. Edit Detection
• 4e. Metadata Examination
• 4f. ENF Analysis
• 4g. Domain-Specific Analysis (TDOA or Enhancement + Transcription)

5. Opinion

Supported determination. Cite only findings from Section 4.

Appendix: Screenshots

Number all screenshots and reference them from Section 4.

Inter-Examiner Comparison⌗

After all individual reports are submitted, your team meets to compare findings. One team member writes and submits the Examiner Comparison Memo — a short document covering:

• Where all four examiners agreed
• Where they diverged, and on what specific points
• How (or whether) the team resolved disagreements
• What the disagreements reveal about the difficulty or ambiguity of the evidence

Final Exam Day: Presentations⌗

Each team presents for approximately 15–20 minutes, structured as follows:

Part 1 — What We Built (5 min)

Shooting teams:

• Scenario overview and case brief summary
• Which recordings were authentic and which fabricated
• Edits applied: what, where, how you tried to conceal them
• Authentication elements embedded
• Reveal your answer key live

CVR teams:

• Scenario overview and case brief summary
• What the degradation consisted of and why you chose it
• What the correct interpretation of the contested claim is
• Reveal your answer key live

Part 2 — What We Found (8 min)

• Walk through your analysis of the opposing team’s case
• Key findings by method
• Final determination and confidence level
• Opposing team responds briefly

Part 3 — Inter-Examiner Comparison (5 min)

• Where did your four examiners agree?
• Where did they diverge, and why?
• What does that tell you about the evidence?

Requirements: Slides required. Analysis screenshots must be included. Every team member must speak.

Submission Checklist⌗

Thursday of Week 15 (before leaving class) — one submission per team member:

• Team_[A or B]_Evidence folder submitted to D2L
  • Audio file(s) in WAV format
  • case_brief.pdf or .docx
• Answer key submitted to D2L separately

Thursday of Week 16 (before team comparison begins) — individual:

• Individual forensic report .docx submitted to D2L

Thursday of Week 16 (before leaving class) — one submission per team:

• Examiner Comparison Memo submitted to D2L

Grading Rubric⌗

Case Building — Shooting Team (25 pts — team grade)⌗

Case Building — CVR Team (25 pts — team grade)⌗

Individual Analysis Report (40 pts — individual grade)⌗

Presentation (35 pts — team grade)⌗

Resources⌗

• Week 5 — Splices and edit detection
• Week 6 — ENF analysis and metadata consistency
• Week 8 — Audio enhancement techniques
• Week 11 — Acoustic event analysis
• Week 12 — Physics of firearm sound signatures
• Week 14 — Cockpit Voice Recorders