Welcome to the official documentation for the Sonodit Audio Pipeline. Our platform processes raw voiceover files and transforms them into broadcast-ready, mastered assets in a 100% autonomous workflow. Below is the chronological journey your audio takes within our intelligent engine:

Step 1: Ingest and Acoustic Profiling

As soon as the file enters the system, the 'Sonodit Audio Profiler' performs a full diagnostic of the signal. It measures the sample rate, bit depth, and calculates the initial Crest Factor. Simultaneously, a neural network analyzes the speaker's phonetic map to determine their precise Words Per Minute (WPM), native language, and regional dialect. This step generates a predictive roadmap that guides the rest of the toolkit.

Step 2: Subsonic Filtering and Noise Subtraction

To ensure maximum clarity, the audio enters the restoration phase. A dynamic High-Pass Filter set at 50 Hz is applied to eradicate parasitic frequencies that clutter the low end. At the same time, the engine samples the 'noise DNA' detected during the speaker's pauses and applies an inverse spectral subtraction algorithm at 50%, removing room hiss without altering the natural timbre of the voice.

Step 3: Block Stabilization (Clip Gain Matching)

One of the most complex phases: the audio undergoes double normalization. First, the entire signal is standardized to -25.0 dB RMS. Immediately after, the system analyzes the audio phrase by phrase. If the voiceover artist whispered a line or shouted a slogan, the algorithm calculates the exact volume difference and corrects it independently, applying a +/-15.0 dB safety limit while protecting short pauses through an exclusion window. This ensures every word has consistent presence and commercial impact.

Step 4: Dynamic Isolation (Predictive Noise Gate)

The stabilized audio passes through an ultra-precise noise gate set with a threshold between -50 dBFS and -55 dBFS. The system detects the voice's attack transients to open the signal flow instantly and applies a controlled fade-out on the tail ends of words, burying any residual air or unwanted breathing in absolute silence.

Step 5: Acoustic Modeling and Commercial Texturizing

The cleaned file is automatically injected into our virtualized hardware engine (DSP Engine). Here, the audio is processed through a simulated analog rack that applies soft-knee compression, strict sibilance control (De-esser), and parallel harmonic excitation. This phase glues the signal together and gives the voice the classic 'big and clear' sound characteristic of major radio networks and cinematic productions, strictly capping the maximum peak at -2.0 dB True Peak.

Step 6: Precision Segmentation and PTK Shield

In the final stage, the temporal orchestrator (Cartographer) takes control. If the system is working autonomously, it locates the silence gaps and makes perfect cuts to export phrases into individual files. Each clip includes a 10-millisecond preemptive attack margin (PTK Protection), ensuring that no initial consonants are clipped. If the user requests a unified master file, the system assembles the phrases by injecting dynamic silences calculated based on the speaker's rhythm, resulting in a compact, rhythmic piece ready for broadcast.