12. Audio Production Pipeline

1. Content

How Project Schwarzwald produces its sound: the categories of audio, the tools and the responsibility of each, the workflows, and the separation between offline production and the game engine's runtime audio.

Status: adopted working architecture; individual automation integrations remain subject to validation.

1.1 Purpose and Creative Strategy

Project Schwarzwald is a primarily solo-developed game. Competing directly with AAA studios in animation volume, graphical complexity, gameplay breadth, or production scale is neither realistic nor the project's objective.

Instead, Schwarzwald will concentrate its resources on areas where a smaller developer can achieve exceptional results:

• Lore and written narrative.
• Environmental storytelling.
• Psychological atmosphere.
• Sound design and acoustic immersion.
• A coherent artistic identity.

Audio is a core design pillar, not a finishing detail.

The objective is to make environments, discoveries, machinery, characters, and narrative events memorable through sound, even when visual or gameplay production cannot match the scale of a major studio.

The production philosophy is to standardize a small number of tools, assign each a clear responsibility, and automate repeatable work through AI agents.

Experimentation is appropriate during initial pipeline discovery. Once a suitable architecture is established, production should become repeatable rather than requiring a new technical investigation for every asset.

1.2 Audio Production Categories

The project adopts six categories to organize its audio production.

These categories describe production responsibilities, not separate playback systems.

All categories share the same central production environment, but each has a distinct creation workflow.

1.3 Standard Toolchain and Responsibilities

…

1.3.1 MuseScore Studio — Score-Based Composition

MuseScore Studio is the primary environment for creating and editing structured musical compositions.

Its responsibilities include:

• Musical notes and rhythmic structures.
• Melodies and harmonies.
• Instrumentation and orchestration.
• Recurring motifs and thematic variations.
• Editable scores using formats such as MusicXML and MIDI.

MuseScore will primarily serve compositions that require a recognizable musical identity.

Examples include the main Schwarzwald theme, character themes, significant location themes, recurring leitmotifs, and distinctive musical stingers.

AI agents such as Codex and Claude will generate and modify musical material through MCP integrations and supported file formats.

Initial validation: Codex has already connected to MuseScore through MCP and successfully created a playable, editable melody.

This confirms the basic technical feasibility of agent-assisted composition. The artistic quality of a complete orchestral production remains to be validated.

1.3.2 Ardour — Central Audio Production Workstation

Ardour is the project's standard Digital Audio Workstation (DAW).

It serves as the central environment for turning compositions, generated audio, and source recordings into finished assets.

Its responsibilities include:

• MIDI composition and editing.
• Virtual instruments and orchestral playback.
• Multitrack audio editing.
• Layering and combining sound effects.
• Equalization, compression, and creative processing.
• Arrangement, mixing, and audio export.
• Producing project-specific variants of source recordings.

Ardour is also the primary environment for transforming commercial sound-library recordings into original Schwarzwald effects.

Why Ardour instead of REAPER?

Both programs provide serious professional audio-production capabilities. However, Ardour better matches the project's preference for open-source software, Linux compatibility, and avoiding unnecessary proprietary dependencies.

REAPER remains a capable alternative but is not part of the standard production pipeline.

The practical reliability and breadth of Ardour's agent-control workflow must still be tested.

1.3.3 Audacity — Auxiliary Audio Utility

Audacity is retained for occasional, isolated audio operations.

These include quick inspections, trimming, cleanup, conversions, and simple file adjustments.

It is not intended to become a second production workstation.

Whenever practical, full production and mixing work should remain inside Ardour to avoid unnecessary duplication of tools and projects.

1.3.4 Suno — Contextual Music Generation

Suno provides a fast method of generating atmospheric and situational music.

It is particularly useful for background compositions associated with exploration, sadness, suspense, reflection, or other environmental moods.

Its advantage is efficiency: a solo developer does not have to orchestrate every musical passage manually.

However, a good generated track is not necessarily a distinctive or recognizable track.

Maintaining the exact musical identity of a generated composition across revisions can also be difficult.

Therefore, Suno is not the primary tool for establishing Schwarzwald's signature musical identity.

1.3.5 ElevenLabs — Voice and Targeted Sound Generation

ElevenLabs is the primary source for synthetic voice performances.

It will be used for:

• Character dialogue.
• Selected AudioLogs.
• Narration when appropriate.
• Screams, breathing, and other vocal performances.
• Specific sound effects that are difficult to source or reproduce.

Initial experiments with mechanical sounds and firearms have produced convincing results.

ElevenLabs may therefore be used for high-quality Physical Events when it offers a clear advantage over existing library material.

The main limitation is cost. Credits should be reserved for assets where the service provides meaningful value.

Generated audio may enter the game directly when its quality is already sufficient. Additional processing in Ardour remains optional.

1.3.6 Professional Sound Libraries — Reusable Source Inventory

Purchased professional sound collections and suitably licensed free libraries constitute the project's reusable source inventory.

This includes material acquired from A Sound Effect and collections such as Sonniss GDC releases, subject to their applicable licenses.

These libraries provide a large number of recordings without requiring new AI generations for every effect.

Project convention: Commercial stock-library sounds should be transformed, layered, or otherwise adapted before inclusion in the game.

This is a creative requirement intended to establish sonic consistency and originality. It does not imply that every commercial license legally requires modifications.

1.4 Music Strategy — Identity Versus Context

Schwarzwald deliberately separates its music into two production modes.

1.4.1 Identity-Critical Music

Identity-critical music includes the main theme, recurring motifs, significant character or location themes, and distinctive musical stingers.

These compositions require deliberate artistic control.

The standard workflow is:

Creative Direction → Codex/Claude → MuseScore → Ardour → Final Music Asset

The agent develops or revises melodies, harmonies, rhythms, and instrumentation.

MuseScore preserves the composition in an editable form.

Ardour handles instrument playback, sound production, arrangement, mixing, and export.

A central composition can then be transformed into multiple versions without losing its recognizable musical identity.

For example, a recurring Schwarzwald theme might appear as a full orchestral composition, a minimal piano arrangement, a subdued ambient variation, or a dissonant interpretation during a disturbing discovery.

The shared musical motif establishes continuity between otherwise different scenes.

The main theme's identity, structure and acceptance criteria: See: Vision › 13. Main Theme Musical Identity.

1.4.2 Contextual Music

Contextual music serves a scene's emotional or atmospheric requirements without necessarily functioning as a recognizable theme.

Examples include reflective exploration music, melancholy passages, and general suspense.

The standard workflow is:

Creative Direction → Suno → Optional Ardour Refinement → Final Music Asset

Suno handles the initial composition and production.

Ardour is used when timing, arrangement, mixing, or other adjustments are required.

This avoids spending extensive composition time on music that does not need a unique thematic identity.

1.4.3 Creative Direction and Agent Responsibility

The developer acts as the creative director.

Responsibilities include defining the intended emotional effect, dramatic context, artistic identity, and acceptance criteria.

AI agents handle technical composition and revision.

This approach does not require the developer to master years of music theory before producing original scores.

However, technical fluency with musical notation does not guarantee that an agent will produce a memorable composition.

The workflow must remain iterative: direct, generate, listen, critique, revise, and approve.

1.5 Voice Budget

Both formats belong to the same logging system. They are not separate technologies invented solely to justify production costs.

This distinction simultaneously supports believable worldbuilding and economical production.

The project can contain extensive written lore while reserving ElevenLabs credits for the recordings where voice acting matters most.

The objective is not to maximize the quantity of AudioLogs, but to make the selected recordings memorable.

The installation's TextLogs and AudioLogs: See: Factions › Project Schwarzwald.

1.6 Standard Production Workflows

…

1.6.1 Voice

1. Write and approve the dialogue or log text.
2. Generate the performance using ElevenLabs.
3. Evaluate voice, accent, delivery, and emotional consistency.
4. Process in Ardour when necessary.
5. Use Audacity only for isolated cleanup when appropriate.
6. Export the approved asset.

1.6.2 Music — Identity-Critical

1. Establish narrative purpose and musical direction.
2. Compose or revise through Codex/Claude and MuseScore.
3. Preserve the editable score and MIDI.
4. Produce the composition using virtual instruments and Ardour.
5. Review melodic identity, arrangement, and mixing.
6. Export approved compositions and variations.

1.6.3 Music — Contextual

1. Define the required mood and scene.
2. Generate music in Suno.
3. Audition and select suitable results.
4. Refine in Ardour when necessary.
5. Export the approved asset.

1.6.4 Physical Events

1. Search existing professional sound libraries first.
2. Use ElevenLabs when a suitable source is unavailable or generation offers a clear advantage.
3. Transform and layer stock recordings in Ardour.
4. Evaluate timing, impact, realism, and consistency.
5. Export the approved effect.

ElevenLabs-generated effects may bypass editing if they already satisfy the requirements.

1.6.5 Ambience

1. Select environmental recordings from libraries or ElevenLabs.
2. Combine and balance layers in Ardour.
3. Prepare loops and suitable variations.
4. Check continuity, unwanted repetition, and sound masking.
5. Export approved environmental beds or individual layers.

Dynamic responses to player movement and game events remain the engine's responsibility.

1.6.6 Stingers

1. Determine whether the cue is primarily musical or sound-design-based.
2. Compose distinctive musical cues in MuseScore or source textural cues from libraries/ElevenLabs.
3. Refine the result in Ardour.
4. Evaluate duration, emotional impact, and thematic consistency.
5. Export the approved cue.

1.6.7 UI Sounds

1. Select appropriate source recordings or synthesize sounds in Ardour.
2. Establish consistent duration and tonal character.
3. Refine timing, clarity, and relative loudness.
4. Export an approved UI sound set.

1.7 Production Audio and Runtime 3D Audio

Offline sound production and runtime spatial audio are separate architectural responsibilities.

MuseScore, Ardour, Audacity, Suno, and ElevenLabs exist to create or process audio assets.

They are not responsible for simulating the acoustic behavior of the game world.

1.7.1 Offline Production

Ardour handles:

• Source-sound design.
• Creative effects and timbral processing.
• Layering and sound construction.
• Music production and mixing.
• Preparation and export of audio files.

1.7.2 Runtime Audio — Game Engine

The Game Engine handles:

• Three-dimensional sound positioning.
• Distance attenuation.
• Binaural/HRTF rendering where applicable.
• Sound occlusion and diffraction.
• Environmental reflections and reverberation.
• Sound propagation through rooms and corridors.
• Dynamic audio behavior based on world state.

For example, a firearm discharge may use the same core source recording in different locations.

Its perceived acoustic characteristics should change according to the physical environment, rather than requiring a separate baked recording for every room.

Where practical, positional sounds should be exported without irreversible, environment-specific spatial processing.

Creative sound processing that defines the source itself remains part of offline production.

The game engine alone owns the real-time acoustic simulation.

Achieving advanced spatial sound comparable to compelling high-end game-audio experiences is a separate technical objective, not a responsibility assigned to Ardour.

1.8 AI Automation and Operational Discipline

The overall production model is:

Human Direction → AI Execution → Audible Review → Targeted Revision → Approved Artifact

Codex and Claude are intended to operate MuseScore and Ardour through MCP integrations or scripts wherever feasible.

The goal is not to eliminate artistic direction.

Instead, agents should handle technical implementation while the developer controls the intended experience.

Each approved asset should preserve, where applicable:

• The original recording or generation reference.
• Editable scores and MIDI.
• Ardour projects or processing configurations.
• Approved rendered audio files.
• Relevant source and licensing information.

This ensures that important assets remain revisable rather than becoming isolated, unrepeatable generations.

The project should prefer one established production path per task.

New tools should only enter the pipeline when they provide a clearly necessary capability unavailable in the existing toolchain.

1.9 Budget, Originality, and Quality Control

The pipeline is designed around the financial realities of solo development.

Existing sound libraries should be the default source for routine physical and environmental effects.

ElevenLabs should be reserved for voice performances and specific sound effects that justify its cost.

Suno should handle contextual music efficiently.

Custom composition effort should be concentrated on music that establishes Schwarzwald's identity.

Stock recordings must be adapted according to the project's originality convention, while generated ElevenLabs audio may be accepted directly when suitable.

Every asset must still be reviewed by listening.

Automation can reduce production effort, but it does not guarantee quality.

Commercial-use licenses and applicable limitations for generated music, voices, sound effects, and sound libraries must be verified before release.

Original source libraries and license records should remain separate from game-distributable assets.

Reference music should inform artistic direction, not become a template for reproducing copyrighted compositions.

1.10 Final Architecture

                 Creative Direction
                         |
                    Codex / Claude
                         |
          +--------------+--------------+
          |              |              |
     MuseScore       Suno /         Professional
      Studio       ElevenLabs      Sound Libraries
          |              |              |
          +--------------+--------------+
                         |
                       Ardour
                 Central Production
             Editing / Instruments
               Layering / Mixing
                    Export
                         |
                 Approved Audio
                      Assets
                         |
                    Game Engine
                         |
               Runtime Audio System
              3D Positioning / HRTF
           Occlusion / Sound Propagation
          Reflections / Spatial Reverberation


       Audacity: Auxiliary Cleanup and Conversion

1.10.1 Adopted Decisions

MuseScore Studio: Structured original composition and recurring musical identity.

Ardour: Central environment for audio editing, sound design, instruments, mixing, and production.

Audacity: Auxiliary utility for occasional simple operations.

Suno: Efficient generation of contextual music.

ElevenLabs: Character voices, selected AudioLogs, and high-quality targeted sound effects.

Professional Sound Libraries: Reusable source inventory for physical events, ambience, stingers, and UI sounds.

Game Engine: Exclusive responsibility for runtime spatial audio and environmental acoustics.

1.10.2 Validated Capabilities

The basic Codex–MCP–MuseScore composition workflow has been demonstrated successfully.

Initial ElevenLabs experiments with voices and sound effects have produced promising results.

Suno has demonstrated sufficient quality for contextual music.

1.10.3 Remaining Validation

The following areas still require practical verification:

• Reliability and scope of AI-agent control over Ardour.
• Orchestral quality achievable through the complete MuseScore-to-Ardour production process.
• Project-wide standards for loudness, file formats, and exporting.
• The game engine's spatial audio architecture and implementation.

1.11 Final Production Principle

Schwarzwald will not attempt to reproduce the production scale of a AAA studio.

Instead, it will pursue exceptional atmosphere, narrative depth, musical identity, and acoustic immersion using a focused set of professional tools and AI-assisted production.

The objective is not to maximize the number of applications, effects, or generated tracks.

The objective is to establish a consistent, distinctive, economical, and reproducible audio pipeline that one developer can operate with the assistance of AI agents.

Creative direction remains human. Technical execution becomes increasingly automated. Audio identity remains specific to Schwarzwald.