Trixy7/trixy_core/audio_morpher/morpher.py

# -*- coding: utf-8 -*-
"""
AudioMorpher — Zentrale Klasse fuer Voice-Morphing.

Verwaltet Presets und wendet DSP-Effekte in definierter Reihenfolge an.
"""

from __future__ import annotations

from pathlib import Path

import numpy as np

from trixy_core.audio_morpher.presets import VoiceMorphPreset, load_presets_from_directory
from trixy_core.audio_morpher.dsp import (
    pitch_shift, speed_change, vtlp, add_breathiness, add_shimmer, apply_fade,
)
from trixy_core.audio_morpher.dynamics import (
    loud_pitch_boost, scale_pauses, loud_tempo_scale,
)
from trixy_core.audio_morpher.effects import (
    apply_reverb, apply_lowpass, apply_highpass, apply_flanger, apply_distortion,
)

# Standard-Preset-Verzeichnis
_DEFAULT_PRESET_DIR = Path(__file__).resolve().parents[2] / "assets" / "default" / "audio_morpher"


class AudioMorpher:
    """
    Voice-Morphing mit Preset-Verwaltung.

    Laedt Presets aus JSON-Dateien und wendet DSP-Effekte in
    definierter Reihenfolge an.
    """

    def __init__(self, preset_dir: Path | str | None = None) -> None:
        """Laedt alle .json-Dateien im Verzeichnis."""
        self._preset_dir = Path(preset_dir) if preset_dir else _DEFAULT_PRESET_DIR
        self._presets: list[VoiceMorphPreset] = []
        self._preset_map: dict[str, VoiceMorphPreset] = {}
        self.reload_presets()

    @property
    def presets(self) -> list[VoiceMorphPreset]:
        """Alle geladenen Presets."""
        return self._presets

    @property
    def preset_ids(self) -> list[str]:
        """IDs aller geladenen Presets."""
        return [p.preset_id for p in self._presets]

    def get_preset(self, preset_id: str) -> VoiceMorphPreset | None:
        """Gibt Preset anhand der ID zurueck oder None."""
        return self._preset_map.get(preset_id)

    def reload_presets(self, preset_dir: Path | str | None = None) -> None:
        """Laedt Presets neu aus dem Verzeichnis."""
        if preset_dir is not None:
            self._preset_dir = Path(preset_dir)
        self._presets = load_presets_from_directory(self._preset_dir)
        self._preset_map = {p.preset_id: p for p in self._presets}

    def morph(
        self,
        audio: np.ndarray,
        sample_rate: int,
        preset: VoiceMorphPreset | str,
        intensity: float = 1.0,
    ) -> np.ndarray:
        """
        Wendet Voice-Morphing auf Audio an.

        Reihenfolge:
        1. Dynamik (loud_pitch_boost, scale_pauses, loud_tempo_scale)
        2. Speed
        3. Pitch-Shift
        4. VTLP (formant_extra)
        5. Breathiness
        6. Shimmer
        """
        if isinstance(preset, str):
            resolved = self._preset_map.get(preset)
            if resolved is None:
                return audio
            preset = resolved

        if intensity < 0.01:
            return audio

        audio = audio.astype(np.float64)

        # Effekt-Parameter mit Intensitaet skalieren
        semitones = preset.pitch_semitones * intensity
        spd = 1.0 + (preset.speed_factor - 1.0) * intensity
        formant = preset.formant_extra * intensity
        breath = preset.breathiness * intensity
        shim = preset.shimmer * intensity
        lpb = preset.loud_pitch_boost * intensity
        ps = 1.0 + (preset.pause_scale - 1.0) * intensity
        lts = 1.0 + (preset.loud_tempo_scale - 1.0) * intensity

        # 1. Dynamik-Effekte
        if abs(lpb) > 0.05:
            audio = loud_pitch_boost(audio, sample_rate, lpb)

        if abs(ps - 1.0) > 0.005:
            audio = scale_pauses(audio, sample_rate, ps)

        if abs(lts - 1.0) > 0.005:
            audio = loud_tempo_scale(audio, sample_rate, lts)

        # 2. Speed
        if abs(spd - 1.0) > 0.005:
            audio = speed_change(audio, sample_rate, spd)

        # 3. Pitch-Shift
        if abs(semitones) > 0.1:
            audio = pitch_shift(audio, sample_rate, semitones)

        # 4. VTLP (Formant-Verschiebung)
        if abs(formant) > 0.01:
            alpha = 1.0 + formant
            audio = vtlp(audio, sample_rate, alpha)

        # 5. Breathiness
        if breath > 0.005:
            audio = add_breathiness(audio, sample_rate, breath)

        # 6. Shimmer
        if shim > 0.005:
            audio = add_shimmer(audio, sample_rate, shim)

        # 7. Filter (Lowpass / Highpass)
        if preset.lowpass_cutoff > 10.0:
            audio = apply_lowpass(audio, sample_rate, preset.lowpass_cutoff, preset.lowpass_order)

        if preset.highpass_cutoff > 10.0:
            audio = apply_highpass(audio, sample_rate, preset.highpass_cutoff, preset.highpass_order)

        # 8. Distortion
        if preset.distortion_gain > 1.001:
            audio = apply_distortion(audio, sample_rate, preset.distortion_gain, preset.distortion_threshold)

        # 9. Flanger
        if preset.flanger_wet > 0.005:
            audio = apply_flanger(audio, sample_rate, preset.flanger_rate, preset.flanger_depth, preset.flanger_wet)

        # 10. Reverb
        if preset.reverb_wet > 0.005:
            audio = apply_reverb(audio, sample_rate, preset.reverb_room_size, preset.reverb_damping, preset.reverb_wet)

        # 11. Trim — Anfang/Ende abschneiden (z.B. VTLP-Artefakte)
        ltrim_samples = int(sample_rate * preset.ltrim) if preset.ltrim > 0.001 else 0
        rtrim_samples = int(sample_rate * preset.rtrim) if preset.rtrim > 0.001 else 0
        if ltrim_samples + rtrim_samples > 0 and ltrim_samples + rtrim_samples < len(audio):
            end = len(audio) - rtrim_samples if rtrim_samples > 0 else len(audio)
            audio = audio[ltrim_samples:end]

        # 8. Fade-In/Out gegen Klick-Artefakte
        audio = apply_fade(audio, sample_rate)

        return audio


def apply_voice_morph(
    audio: np.ndarray,
    sample_rate: int,
    preset: VoiceMorphPreset | str,
    intensity: float = 1.0,
) -> np.ndarray:
    """Convenience-Funktion — erstellt AudioMorpher und wendet Preset an."""
    morpher = AudioMorpher()
    return morpher.morph(audio, sample_rate, preset, intensity)