Audio Production Applications

Introduction

Audio is a central medium in radio, podcasts, advertising, film, streaming, social media and branded communication. Although modern software has made recording and editing more accessible, professional audio production still depends on informed decisions about acoustics, microphones, digital sampling, editing, mixing, loudness, metadata and distribution. Poorly recorded audio cannot always be repaired in post-production; conversely, a technically clean recording may still fail if its structure, tone or message is inappropriate for its audience.

Audio production can be understood as a managed signal chain:

plan → capture → edit → mix → master → export → quality-control → distribute

Each stage affects later stages. For example, excessive recording levels may cause irreversible clipping, while an unsuitable sample rate may complicate later video or broadcast integration. The media professional therefore needs both creative judgment and technical literacy.

Learning Objectives

By the end of this chapter, students should be able to:

1. Plan an audio project according to audience, platform, content and production resources.
2. Select microphones, recording environments and digital settings appropriately.
3. Explain sampling rate, bit depth, digital clipping and file formats.
4. Record dialogue, interviews, music and advertisements using effective gain staging.
5. Edit spoken-word audio non-destructively and ethically.
6. Apply equalization, compression, gating, de-essing, noise reduction and limiting appropriately.
7. Mix audio for intelligibility, balance and emotional effect.
8. Master and export audio using suitable loudness, peak and delivery standards.
9. Diagnose common technical problems and evaluate the limitations of restoration and AI tools.

1. Planning the Production

1.1 Define the brief

An audio brief should identify:

• Purpose: information, entertainment, persuasion, documentation or brand communication.
• Audience: age, language, listening context and accessibility needs.
• Format: podcast episode, radio spot, interview, documentary, music bed, audiobook or social-media clip.
• Duration: total running time and any segment limits.
• Platform: broadcast system, podcast host, website, streaming service or social network.
• Delivery requirements: file format, sample rate, loudness target, naming convention, metadata and deadline.
• Rights and permissions: music licences, performer consent, location permissions and copyright clearance.

A production plan should also identify risks. Outdoor recordings may be affected by wind and traffic; remote interviews may suffer from unstable internet connections; advertisements may require multiple language versions and legal disclaimers.

1.2 Pre-production documents

A script or running order establishes the sequence of speech, music, ambience, effects and silence. A simple production log records take numbers, microphone positions, technical settings and problems. A release form documents permission to record and publish a person’s voice. These documents are not administrative extras: they protect editorial accuracy and reduce expensive rework.

For a podcast interview, pre-production may include researching the guest, preparing open-ended questions, testing remote recording, obtaining consent, and agreeing on whether editing may remove pauses or repeated statements. For a radio commercial, it may include timing the copy, selecting music, recording several voice styles and checking mandatory disclosures.

2. Sound and Digital Audio Fundamentals

2.1 Sound as a physical phenomenon

Sound is a variation in air pressure travelling through a medium. Frequency, measured in hertz (Hz), describes cycles per second and is associated with perceived pitch. Amplitude describes the magnitude of pressure variation and is related to perceived level. Timbre is the characteristic quality that enables listeners to distinguish instruments or voices with similar pitch and loudness.

Microphones convert acoustic pressure into electrical signals. An audio interface then converts the analogue signal into numerical samples that software can process. During playback, a digital-to-analogue converter reconstructs an electrical waveform, which is sent to speakers or headphones.

2.2 Sample rate

The sample rate is the number of amplitude measurements made each second. It is expressed in samples per second, commonly 44.1 kHz or 48 kHz. According to the Nyquist–Shannon sampling principle, a band-limited signal can be reconstructed when the sampling frequency is greater than twice its highest frequency component:

Here, f_s is the sampling frequency in hertz and f_max is the highest frequency in the signal, also in hertz. The condition assumes that the signal is adequately band-limited before sampling. Frequencies above the permitted range can produce aliasing, in which unwanted components appear at lower frequencies. An anti-aliasing filter is therefore used before conversion.

Human hearing is often described as extending approximately from 20 Hz to 20 kHz, although the practical range varies with age and listening conditions. A 44.1 kHz sample rate supports frequencies below 22.05 kHz in the ideal mathematical limit; a 48 kHz rate supports frequencies below 24 kHz. In video production, 48 kHz is the usual production choice because it aligns with common video and broadcast workflows. For music distribution, 44.1 kHz remains common. The best practice is to select the rate required by the final delivery system and avoid unnecessary conversion.

2.3 Bit depth and dynamic range

Bit depth determines how precisely each sample’s amplitude is represented. Increasing bit depth reduces quantization error and increases theoretical dynamic range. For ideal linear PCM audio, the approximate theoretical dynamic range is:

DR is theoretical dynamic range in decibels, and N is bit depth in bits per sample. The equation assumes ideal quantization and does not include analogue circuitry, microphone self-noise or converter imperfections. At 16-bit depth, the theoretical value is approximately 98 dB; at 24-bit depth, it is approximately 146 dB. Real equipment achieves less because of electronic noise.

24-bit recording is generally appropriate for production because it allows comfortable headroom without requiring recording levels close to digital full scale. Bit depth does not make a quiet microphone louder; it provides finer numerical resolution and greater tolerance during capture.

2.4 Decibels and digital levels

The decibel is a logarithmic ratio. For power quantities:

For amplitude quantities such as voltage, when the impedance is the same:

P₁ and P₂ are power values, while A₁ and A₂ are amplitude values. The reference must be specified. In digital audio, dBFS means decibels relative to full scale; 0 dBFS is the maximum representable peak, and normal sample values are negative. A signal exceeding 0 dBFS clips in a fixed-point digital system.

Peak level and loudness are different. A short transient can have a high peak but contribute relatively little perceived loudness. Loudness measurement therefore uses time-weighted methods such as LUFS (Loudness Units relative to Full Scale), usually with integrated, short-term and momentary readings. Loudness standards vary by platform and jurisdiction; a producer must consult the current delivery specification rather than treating one target as universal.

3. Equipment and Recording Practice

3.1 Microphone selection

Microphone choice depends on source, distance, room and desired character.

Polar patterns describe directional sensitivity. Cardioid microphones reject much sound from the rear, while omnidirectional microphones capture sound broadly and may provide natural low-frequency response when placed close to a source. Directionality is not a substitute for a quiet room: reflections and noise can still enter through the microphone’s sensitive region.

3.2 Placement and room control

Microphone distance strongly affects direct sound, room sound and proximity effect. A close voice recording may sound intimate but can exaggerate plosive consonants and low frequencies. A distant recording contains more reverberation and background noise. The microphone should normally be positioned slightly off-axis to reduce plosives and harsh consonants, with a pop filter and stable stand.

Room treatment aims to control reflections rather than create complete silence. Soft furnishings, absorptive panels and suitable microphone placement can reduce early reflections. Portable isolation shields may help in some situations but cannot correct severe room echo. Headphones should be used to monitor noise, clipping, hum, clothing movement and microphone handling.

3.3 Gain staging

Gain staging means maintaining appropriate signal levels through the complete chain. Set the microphone preamp high enough for a healthy recording but leave headroom for unexpected peaks. Do not use the master fader to repair an input signal that was recorded too quietly or clipped at the preamp.

A practical 24-bit dialogue recording may average substantially below 0 dBFS and peak around −12 to −6 dBFS, depending on the source and workflow. These values are working guidelines, not universal laws. The essential rule is to avoid clipping while keeping the noise floor acceptably low. Record a test passage at the loudest expected performance level, monitor with headphones, and check both meters and the actual sound.

4. Digital Audio Workstations and Non-Destructive Editing

A digital audio workstation (DAW) provides tracks, clips, timelines, plug-ins, automation, routing and export functions. Common operations include cutting, moving, trimming, fading, crossfading, adding markers and adjusting clip gain.

Non-destructive editing preserves the original media and applies changes through project instructions or processing chains. Keep an untouched original, a clearly named working session and regular versioned backups. Use descriptive names such as Podcast07_Interview_Edit_v03, rather than overwriting one file repeatedly.

4.1 Spoken-word editing

A professional spoken-word edit should improve clarity and structure without making the speaker sound unnatural. Remove false starts, long irrelevant pauses, repeated words and technical interruptions only when editorially justified. Retain meaningful hesitation, emotion and culturally significant speech patterns where they contribute to authenticity.

Useful practices include:

• Edit at waveform zero crossings when possible, or use short fades to avoid clicks.
• Use room tone to fill abrupt gaps rather than inserting digital silence everywhere.
• Match ambience between edits.
• Avoid cutting breaths mechanically; reduce distracting breaths selectively.
• Use clip gain before compression to correct large level differences.
• Maintain a transcript or edit decision list for important factual content.

Music and effects should be placed on separate tracks so that their levels can be automated independently. Crossfades prevent clicks between clips, while a short fade-in and fade-out prevents accidental start and end transients.

5. Noise, Distortion and Restoration

5.1 Identify the problem before processing

Noise may be broadband hiss, low-frequency rumble, electrical hum, intermittent interference, traffic, air-conditioning or room reverberation. Each problem requires a different response. A high-pass filter can reduce rumble but cannot remove a nearby conversation. A noise gate can lower noise during pauses but may produce unnatural cut-offs and does not improve noisy speech during words.

5.2 Noise reduction

Spectral noise reduction often learns a noise profile from a short section containing only unwanted sound and attenuates similar frequencies throughout the clip. It is most effective when the noise is relatively constant. Excessive reduction causes musical artifacts, watery textures or consonant damage. Apply the smallest amount that solves the editorial problem, and compare processed and unprocessed versions at matched loudness.

Hum often occurs at the mains frequency and its harmonics. A notch filter may reduce it, but removing too much nearby frequency can make speech thin. Re-recording in a better environment is often superior to aggressive restoration.

5.3 Distortion and clipping

Clipping occurs when the signal exceeds the available numerical or analogue range. A clipped waveform contains added harmonic distortion. Restoration tools may estimate missing peaks, but they cannot reliably recover the exact original waveform. If the recording is important, a clean re-take is preferable.

6. Processing and Mixing

6.1 Equalization

Equalization (EQ) changes the level of selected frequency ranges. A high-pass filter may remove subsonic rumble; a narrow cut may reduce resonance; a gentle presence boost may improve intelligibility. EQ should be guided by listening and context rather than frequency recipes. Boosting every voice in the same frequency range can increase harshness and reduce space for music.

Use subtractive EQ before additive EQ where practical. Compare the processed signal with the original at similar loudness, because louder often appears subjectively better even when it is not.

6.2 Compression

A compressor reduces dynamic range when the signal exceeds a threshold. Important controls include threshold, ratio, attack, release and make-up gain. If L_in is the input level above threshold, an idealized static compressor with ratio R produces an output excess approximately:

This simplified relationship applies only to the portion above threshold and assumes a hard-knee model without attack, release or detector complications. A 4:1 ratio means that a 4 dB increase above threshold produces approximately a 1 dB increase at the output above threshold. Real compressors may use soft knees, programme-dependent timing and different detector modes.

Moderate compression can stabilize dialogue and make a commercial sound confident. Excessive compression raises room noise, removes natural emphasis and causes pumping. Attack and release should respect the source: very fast attack can reduce consonant impact, while release that is too fast can create audible level modulation.

6.3 De-essing and gating

A de-esser is a frequency-conscious compressor that reduces excessive sibilance, commonly in an upper-mid or high-frequency region. It should act mainly on problematic consonants, not dull the entire voice.

A gate or downward expander reduces level below a threshold. It can control headphone bleed or pauses in a controlled recording, but it should not be used to hide poor room tone without careful timing. Manual clip gain and room-tone editing often sound more natural for dialogue.

6.4 Balancing the mix

Mixing establishes hierarchy. In a podcast, speech normally remains intelligible above music and ambience. In an advertisement, the voice, brand message and legal copy must remain understandable on small speakers and in noisy environments. Use automation rather than relying only on static levels. Ducking music beneath speech can be achieved with manual automation or side-chain compression, but the transitions should be smooth and musically appropriate.

Panning creates spatial separation, but spoken-word productions are often kept centred for compatibility. Check mono compatibility, especially when using stereo widening or phase-based effects. Excessive width may disappear or change character when summed to mono.

7. Loudness, Mastering and Delivery

Mastering is the final technical and aesthetic preparation of a mix for a specified destination. It may include broad tonal correction, gentle bus compression, limiting, sequencing, metadata preparation and quality control. Mastering cannot compensate for a fundamentally unbalanced mix.

7.1 Loudness management

Peak meters show instantaneous sample or reconstructed peaks; loudness meters estimate perceived level over time. True-peak measurement is valuable because digital-to-analogue conversion and codec processing can create inter-sample peaks above the sample peak. A limiter can prevent excessive peaks, but a low ceiling alone does not guarantee suitable loudness.

There is no single loudness value appropriate for every delivery context. Broadcast specifications, podcast platforms, music services and advertising systems may use different integrated loudness and maximum true-peak requirements. Producers should obtain the current technical specification and document the target in the project notes. If no specification is supplied, choose a conservative, platform-appropriate target and prioritize intelligibility and dynamic integrity over competitive loudness.

7.2 Export formats

WAV or another uncompressed PCM format is preferred as a production master because it preserves the samples without lossy compression. MP3, AAC and similar formats reduce file size through perceptual coding and are appropriate when required for distribution, but repeated transcoding should be avoided.

Export decisions include:

• sample rate and bit depth;
• mono, stereo or multichannel configuration;
• file format and codec settings;
• integrated loudness and true-peak limits;
• start and end padding;
• metadata such as title, author, episode and copyright;
• file naming and version number.

For a mono voice podcast, exporting two identical stereo channels wastes data and may create confusion. For a music or ambience production, stereo may be necessary. Always audition the exported file, not only the DAW session.

7.3 Quality-control checklist

Before delivery, verify:

1. The file opens and plays from beginning to end.
2. There are no unintended clicks, pops, dropouts or clipped peaks.
3. Dialogue is intelligible on headphones, speakers and a modest mobile device.
4. Music and effects do not obscure words or legal information.
5. Loudness and true-peak measurements meet the delivery brief.
6. The correct sample rate, bit depth, channel format and codec are used.
7. Metadata, spelling, credits, permissions and version numbers are correct.
8. The exported file is stored with the project archive and an appropriate backup.

Worked Example: Estimating Uncompressed File Size

A producer records a 30-minute stereo programme at 48 kHz and 24-bit depth. The approximate uncompressed data size is calculated as:

where S is file size in bytes, t is duration in seconds, f_s is sample rate in samples per second, N is bit depth in bits per sample, and C is the number of channels. The equation assumes uncompressed PCM and excludes headers and metadata.

Here, t = 30 × 60 = 1800 seconds, f_s = 48,000, N/8 = 24/8 = 3 bytes per sample, and C = 2. Therefore:

Using decimal units, this is approximately 518.4 MB. Binary operating systems may display a somewhat smaller value in MiB because 1 MiB equals 1,048,576 bytes. The example illustrates why production masters require more storage than compressed delivery files.

Applied Case: Illustrative Podcast Production

Background

A university media team is producing a 25-minute podcast episode about student entrepreneurship. The episode includes a host, a remote guest interview, a short music introduction and two sponsor messages.

Problem

The guest’s recording contains steady air-conditioner noise, inconsistent distance from the microphone and several long pauses. The sponsor requires clear legal wording, while the distribution service specifies a particular loudness range and file format.

Analysis

The team first separates problems that can be edited from problems that require restoration. Pauses and repeated phrases can be edited manually. The constant air-conditioner noise may be reduced with a carefully captured noise profile, but aggressive processing could damage speech. Level inconsistency can be addressed with clip gain followed by moderate compression. The legal wording must not be shortened in a way that changes meaning.

Approach and solution

The team keeps the original interview file and creates a versioned DAW session. The host and guest are placed on separate tracks. The most distracting noise is reduced conservatively, and room tone is inserted at selected edit points. Clip gain is used to bring quiet guest sections closer to the intended level before compression. A high-pass filter removes low-frequency rumble, while a de-esser controls occasional harsh sibilance.

The music is kept on a separate stereo track and automated downward during speech. Sponsor copy is recorded again in a quieter room rather than relying on heavy restoration. The completed mix is checked in headphones, laptop speakers and a phone. Loudness and true-peak values are measured against the current delivery specification. A high-quality PCM master and the required distribution encode are exported, named and archived.

Outcome

The final episode sounds consistent and intelligible without pretending that the remote recording was made in a studio. The team preserves the guest’s natural delivery, avoids excessive denoising and meets the platform’s technical requirements. The production file includes notes documenting processing choices and permissions.

Learning points

• Pre-production and recording quality reduce post-production costs.
• Restoration should be proportionate to the problem.
• Editorial authenticity and technical polish must be balanced.
• Loudness compliance is a delivery responsibility, not merely a mastering preference.
• Separate masters and distribution files improve quality control and future reuse.

Professional, Ethical and Governance Issues

Audio producers handle personal data, identity and sometimes sensitive testimony. Obtain informed consent, explain intended distribution and secure files appropriately. Do not deceptively edit a person’s statement to reverse its meaning. Maintain an audit trail for factual or investigative productions.

Copyright applies to music, sound effects, archival recordings and sometimes distinctive performances. A publicly available file is not automatically free to reuse. Licences should specify territory, duration, media and commercial use. Attribution requirements must be followed.

Accessibility requires more than good sound. Podcasts and video-based audio content should provide accurate transcripts or captions where appropriate. Identify speakers, represent meaningful non-speech sounds and review automated transcripts for names, numbers and technical terms.

AI-based tools can separate stems, suppress noise, generate transcripts, clone voices and repair timing. They can improve accessibility and assist repetitive work, but they may introduce artifacts, misrepresent speech or create synthetic statements. Voice cloning requires explicit permission and transparent disclosure. AI output should be reviewed by a human, and original recordings should be retained. A producer remains accountable for the published result.

Contemporary Developments and Best Practices

Cloud collaboration allows producers, editors and clients to work across locations, but it creates version-control and confidentiality risks. Use consistent project structures, secure transfer methods and clear ownership of the master session. Immersive and spatial audio are increasingly used in selected installations, games and streaming productions; they require compatible monitoring and delivery specifications and should not be adopted merely as a novelty.

Recommended practice includes:

• Record the cleanest signal possible before applying effects.
• Use 24-bit production recording when supported by the workflow.
• Match sample rate to the principal delivery environment.
• Keep original media, project files, exports and documentation.
• Use meters and critical listening together.
• Compare processing at matched loudness.
• Check mono compatibility and several playback systems.
• Measure the delivered file, not only the session.
• Treat loudness specifications as platform requirements, not universal aesthetic rules.
• Obtain consent and licences before publication.

Challenges and Limitations

Audio software can create the illusion that every problem is repairable. In reality, noise overlapping the same frequencies as speech, severe reverberation, clipped peaks and missing information may not be recoverable without artifacts. Loudness meters also provide measurements rather than judgments; two programmes with similar integrated loudness can sound different because of dynamics, spectral balance and content. Finally, technical perfection cannot replace editorial relevance, compelling storytelling and respect for contributors.

Chapter Summary

Audio production is a complete workflow extending from brief development to final delivery. Effective results depend on appropriate microphones, controlled recording environments, safe gain staging and suitable sample-rate and bit-depth decisions. Sampling theory explains why the sample rate must exceed twice the highest retained frequency, while bit depth affects quantization resolution and theoretical dynamic range.

A DAW supports non-destructive editing, routing, processing and automation. Noise reduction, EQ, compression, de-essing, gating and limiting are useful only when applied for a defined purpose and monitored critically. Mixing establishes intelligibility and hierarchy, while mastering prepares a specified delivery rather than simply making audio louder. Loudness, true peak, file format, metadata and quality control must be checked in the exported file.

Professional practice also includes consent, copyright, accessibility, secure archiving and transparent use of AI. The most reliable production strategy is preventive: plan carefully, record cleanly, preserve originals, edit ethically and validate every deliverable.

Key Terms

Aliasing: Unwanted frequency distortion caused when a signal is sampled below the required rate.

Bit depth: Number of bits used to represent each digital audio sample.

Clipping: Distortion caused when a signal exceeds the available amplitude range.

Compression: Dynamic processing that reduces level differences above a threshold.

DAW: Digital audio workstation used to record, edit, process and export audio.

dBFS: Digital level expressed relative to full scale, where 0 dBFS is the maximum nominal level.

Dynamic range: Difference between the quietest usable and loudest undistorted signal levels.

EQ: Equalization; adjustment of selected frequency regions.

Gain staging: Management of signal level through each part of an audio chain.

LUFS: Loudness Units relative to Full Scale, used for standardized loudness measurement.

Mastering: Final preparation of an audio mix for a defined delivery format or platform.

Noise floor: Background noise level present in a recording system or environment.

Nyquist frequency: Half the sampling rate; the highest ideal frequency representable without aliasing.

Polar pattern: A microphone’s directional sensitivity around its capsule.

Sample rate: Number of digital samples recorded per second.

Side-chain: A control signal used to trigger processing, such as music ducking under speech.

True peak: Estimated peak level including possible inter-sample peaks after conversion.

Waveform: Visual or mathematical representation of changing amplitude over time.

Review Questions

Short Answer Questions

1. What is the difference between sample rate and bit depth?
2. Why is gain staging important during recording?
3. Distinguish peak level from loudness.
4. Give two advantages of non-destructive editing.
5. What is the purpose of a de-esser?

Long Answer Questions

1. Explain the complete audio production workflow for a podcast, from planning through distribution.
2. Discuss how microphone selection, placement and room acoustics affect dialogue quality.
3. Explain sampling, aliasing and bit depth, including the assumptions behind the relevant technical principles.
4. Describe how EQ, compression, noise reduction and limiting can be combined in a spoken-word mix.
5. Discuss ethical, copyright and accessibility responsibilities in professional audio production.

Analytical / Application Questions

1. A field interview has strong wind noise, uneven speech levels and occasional clipping. Identify which problems can be prevented, which can be treated in post-production and which may require a re-recording. Justify your decisions.
2. Design a technical delivery plan for a 30-second radio advertisement and explain how you would verify loudness, true peak, intelligibility, file format and legal copy.
3. A podcast sounds loud in headphones but unclear on a mobile speaker. Analyse possible causes and propose a systematic troubleshooting and revision process.

Multiple Choice Questions with Answers

Question 1

Which condition is required to avoid aliasing for a band-limited signal?

A. The sample rate must be greater than twice the highest frequency component.

B. The bit depth must be greater than the sample rate.

C. The signal must peak at exactly 0 dBFS.

D. The file must be exported as MP3.

Correct Answer: A. The sample rate must be greater than twice the highest frequency component.

Question 2

What does 0 dBFS represent in a conventional fixed-point digital audio system?

A. The average listening level

B. The maximum representable nominal level

C. The microphone’s noise floor

D. The recommended podcast loudness

Correct Answer: B. The maximum representable nominal level

Question 3

Which microphone is commonly suited to discreet dialogue capture on a performer?

A. Lavalier microphone

B. Ribbon microphone placed across the room

C. Boundary microphone used outdoors without protection

D. Small-diaphragm microphone aimed at the ceiling

Correct Answer: A. Lavalier microphone

Question 4

What is the primary purpose of a compressor?

A. To convert mono audio into stereo

B. To reduce dynamic range according to control settings

C. To remove all room reverberation

D. To increase the sample rate

Correct Answer: B. To reduce dynamic range according to control settings

Question 5

Why should a producer retain an untouched original recording?

A. It guarantees that all clipping can be repaired.

B. It permits recovery, re-editing and comparison with later processing.

C. It automatically satisfies every platform loudness standard.

D. It prevents the need for metadata.

Correct Answer: B. It permits recovery, re-editing and comparison with later processing.

Question 6

A 4:1 compressor ratio ideally makes a 4 dB increase above threshold become approximately what increase above threshold?

A. 16 dB

B. 8 dB

C. 4 dB

D. 1 dB

Correct Answer: D. 1 dB

Question 7

Which measurement is specifically associated with perceived programme loudness rather than an instantaneous sample peak?

A. LUFS

B. Hertz

C. Bit depth

D. Frame rate

Correct Answer: A. LUFS

Question 8

Why is a WAV file commonly used as a production master?

A. It is always smaller than a perceptually coded file.

B. It preserves uncompressed PCM audio suitable for further work.

C. It automatically removes background noise.

D. It cannot contain metadata.

Correct Answer: B. It preserves uncompressed PCM audio suitable for further work.

Question 9

What is the most appropriate response to severe clipped dialogue when an original clean take exists?

A. Use the clean take instead of relying on aggressive restoration.

B. Apply more limiting until the clipping disappears.

C. Increase the sample rate after export.

D. Add stereo widening.

Correct Answer: A. Use the clean take instead of relying on aggressive restoration.

Question 10

Which practice best supports ethical use of AI voice tools?

A. Clone any publicly available voice without disclosure.

B. Delete the original recording after synthesis.

C. Obtain explicit permission and review and disclose the generated result appropriately.

D. Use AI output without checking names or factual statements.

Correct Answer: C. Obtain explicit permission and review and disclose the generated result appropriately.