vfp: update generator

SoulMelody · Jan 8, 2025 · 5f53137 · 5f53137
1 parent e26d633
commit 5f53137
Show file tree

Hide file tree

Showing 4 changed files with 77 additions and 54 deletions.
diff --git a/libresvip/plugins/vfp/model.py b/libresvip/plugins/vfp/model.py
@@ -13,15 +13,15 @@ class VOXFactoryNote(BaseModel):
     name: str
     syllable: str
     ticks: float
-    duration: float
     duration_ticks: float = Field(alias="durationTicks")
-    velocity: int
+    duration: float = 0.25
+    velocity: int = 1
     note_type: Optional[str] = Field(None, alias="noteType")
     vibrato_depth: Optional[float] = Field(None, alias="vibratoDepth")
     pre_bend: Optional[float] = Field(None, alias="preBend")
     post_bend: Optional[float] = Field(None, alias="postBend")
     harmonic_ratio: Optional[float] = Field(None, alias="harmonicRatio")
-    pitch_bends: list[float] = Field(alias="pitchBends")
+    pitch_bends: list[float] = Field(default_factory=list, alias="pitchBends")
 
 
 class VOXFactoryMetadata(BaseModel):
@@ -34,18 +34,18 @@ class VOXFactoryMetadata(BaseModel):
 
 
 class VOXFactoryClipBase(BaseModel):
-    name: str
-    start_quarter: float = Field(alias="startQuarter")
-    offset_quarter: float = Field(alias="offsetQuarter")
+    name: str = ""
+    start_quarter: float = Field(0, alias="startQuarter")
+    offset_quarter: float = Field(0, alias="offsetQuarter")
     length: float
-    use_source: bool = Field(alias="useSource")
+    use_source: bool = Field(True, alias="useSource")
     audio_data_key: Optional[str] = Field(None, alias="audioDataKey")
-    audio_data_order: list[str] = Field(alias="audioDataOrder")
-    audio_data_quarter: float = Field(alias="audioDataQuarter")
+    audio_data_order: list[str] = Field(default_factory=list, alias="audioDataOrder")
+    audio_data_quarter: float = Field(0, alias="audioDataQuarter")
     note_bank: dict[str, VOXFactoryNote] = Field(alias="noteBank")
     note_order: list[str] = Field(alias="noteOrder")
-    next_note_index: int = Field(alias="nextNoteIndex")
-    pinned_audio_data_order: list[str] = Field(alias="pinnedAudioDataOrder")
+    next_note_index: int = Field(0, alias="nextNoteIndex")
+    pinned_audio_data_order: list[str] = Field(default_factory=list, alias="pinnedAudioDataOrder")
     metadata: Optional[VOXFactoryMetadata] = None
 
 
@@ -77,18 +77,18 @@ def extract_audio(self, info: ValidationInfo) -> Self:
 
 
 class VOXFactoryAudioViewProperty(BaseModel):
-    view: str
-    colormap: str
-    window: str
-    window_size: int = Field(alias="windowSize")
-    hop_size: int = Field(alias="hopSize")
-    f_min: float = Field(alias="fMin")
-    f_max: None = Field(alias="fMax")
-    level_min: None = Field(alias="levelMin")
-    level_max: None = Field(alias="levelMax")
-    level_scale: str = Field(alias="levelScale")
-    num_bins: int = Field(alias="numBins")
-    bins_per_octave: int = Field(alias="binsPerOctave")
+    view: str = "waveform"
+    colormap: str = "Heated Metal"
+    window: str = "Hann"
+    window_size: int = Field(1024, alias="windowSize")
+    hop_size: int = Field(256, alias="hopSize")
+    f_min: float = Field(27.5, alias="fMin")
+    f_max: Optional[float] = Field(None, alias="fMax")
+    level_min: Optional[float] = Field(None, alias="levelMin")
+    level_max: Optional[float] = Field(None, alias="levelMax")
+    level_scale: str = Field("dB", alias="levelScale")
+    num_bins: int = Field(230, alias="numBins")
+    bins_per_octave: int = Field(24, alias="binsPerOctave")
 
 
 class VOXFactoryDevice(BaseModel):
@@ -100,19 +100,21 @@ class VOXFactoryDevice(BaseModel):
 
 
 class VOXFactoryTrackBase(BaseModel):
-    name: str
+    name: str = ""
     instrument: Optional[str] = None
-    h: int
-    color: str
-    volume: float
-    pan: float
-    solo: bool
-    mute: bool
-    arm: bool
+    h: int = 3
+    color: str = "#7878f1"
+    volume: float = 1.0
+    pan: float = 0.0
+    solo: bool = False
+    mute: bool = False
+    arm: bool = False
     clip_order: list[str] = Field(alias="clipOrder")
-    device_bank: dict[str, VOXFactoryDevice] = Field(alias="deviceBank")
-    device_order: list[str] = Field(alias="deviceOrder")
-    audio_view_property: VOXFactoryAudioViewProperty = Field(alias="audioViewProperty")
+    device_bank: dict[str, VOXFactoryDevice] = Field(default_factory=dict, alias="deviceBank")
+    device_order: list[str] = Field(default_factory=list, alias="deviceOrder")
+    audio_view_property: VOXFactoryAudioViewProperty = Field(
+        default_factory=VOXFactoryAudioViewProperty, alias="audioViewProperty"
+    )
 
 
 class VOXFactoryVocalTrack(VOXFactoryTrackBase):
@@ -150,13 +152,19 @@ class VOXFactoryAudioData(BaseModel):
 
 
 class VOXFactoryProject(BaseModel):
-    version: str
+    version: str = "0.12.0"
     tempo: float
     time_signature: list[int] = Field(alias="timeSignature")
-    project_name: str = Field(alias="projectName")
+    project_name: str = Field("Untitled Project", alias="projectName")
     track_bank: dict[str, VOXFactoryTrack] = Field(alias="trackBank")
-    track_order: list[str] = Field(alias="trackOrder")
-    selected_track_bank: list[str] = Field(alias="selectedTrackBank")
-    selected_clip_bank: list[VOXFactorySelectedClipBankItem] = Field(alias="selectedClipBank")
-    selected_note_bank: list[VOXFactorySelectedNoteBankItem] = Field(alias="selectedNoteBank")
-    audio_data_bank: dict[str, VOXFactoryAudioData] = Field(alias="audioDataBank")
+    track_order: list[str] = Field(default_factory=list, alias="trackOrder")
+    selected_track_bank: list[str] = Field(default_factory=list, alias="selectedTrackBank")
+    selected_clip_bank: list[VOXFactorySelectedClipBankItem] = Field(
+        default_factory=list, alias="selectedClipBank"
+    )
+    selected_note_bank: list[VOXFactorySelectedNoteBankItem] = Field(
+        default_factory=list, alias="selectedNoteBank"
+    )
+    audio_data_bank: dict[str, VOXFactoryAudioData] = Field(
+        default_factory=dict, alias="audioDataBank"
+    )
diff --git a/libresvip/plugins/vfp/vox_factory_converter.py b/libresvip/plugins/vfp/vox_factory_converter.py
@@ -33,10 +33,11 @@ def dump(self, path: pathlib.Path, project: Project, options: OutputOptions) ->
             archive_file.writestr(
                 "project.json",
                 json.dumps(
-                    vox_factory_project.model_dump(mode="json", exclude_none=True, by_alias=True),
+                    vox_factory_project.model_dump(mode="json", by_alias=True),
                     ensure_ascii=False,
                 ),
             )
+            archive_file.mkdir("resources")
             for audio_name, audio_path in generator.audio_paths.items():
                 archive_file.writestr(f"resources/{audio_name}", audio_path.read_bytes())
         path.write_bytes(buffer.getvalue())
diff --git a/libresvip/plugins/vfp/vox_factory_generator.py b/libresvip/plugins/vfp/vox_factory_generator.py
@@ -1,7 +1,10 @@
 import dataclasses
+import math
 import pathlib
+import secrets
 
-from libresvip.core.constants import DEFAULT_BPM
+from libresvip.core.constants import DEFAULT_BPM, TICKS_IN_BEAT
+from libresvip.core.time_sync import TimeSynchronizer
 from libresvip.model.base import Note, Project, SingingTrack, SongTempo, TimeSignature, Track
 
 from .model import (
@@ -17,9 +20,13 @@
 @dataclasses.dataclass
 class VOXFactoryGenerator:
     options: OutputOptions
+    prefix: str = dataclasses.field(init=False)
     audio_paths: dict[str, pathlib.Path] = dataclasses.field(default_factory=dict)
+    synchronizer: TimeSynchronizer = dataclasses.field(init=False)
 
     def generate_project(self, project: Project) -> VOXFactoryProject:
+        self.prefix = secrets.token_hex(5)
+        self.synchronizer = TimeSynchronizer(project.song_tempo_list)
         vox_project = VOXFactoryProject(
             tempo=self.generate_tempo(project.song_tempo_list),
             time_signature=self.generate_time_signature(project.time_signature_list),
@@ -41,10 +48,9 @@ def generate_tracks(self, tracks: list[Track]) -> dict[str, VOXFactoryTrack]:
         track_bank = {}
         for i, track in enumerate(tracks):
             if isinstance(track, SingingTrack):
-                note_list = self.generate_notes(track.note_list)
-                clip_bank = {f"clip_{i}": clip for i, clip in enumerate(note_list)}
+                clip_bank = self.generate_notes(track.note_list)
                 clip_order = sorted(clip_bank.keys())
-                track_bank[str(i)] = VOXFactoryVocalTrack(
+                track_bank[f"{self.prefix}-tr{i}"] = VOXFactoryVocalTrack(
                     clip_bank=clip_bank,
                     clip_order=clip_order,
                 )
@@ -53,18 +59,27 @@ def generate_tracks(self, tracks: list[Track]) -> dict[str, VOXFactoryTrack]:
     def generate_notes(self, notes: list[Note]) -> dict[str, VOXFactoryVocalClip]:
         note_bank = {}
         note_order = []
+        max_ticks = notes[-1].end_pos if notes else 0
+        max_quarter = max_ticks / TICKS_IN_BEAT
         for i, note in enumerate(notes):
-            note_bank[f"note_{i}"] = self.generate_note(note)
-            note_order.append(f"note_{i}")
-        return {
-            "clip": VOXFactoryVocalClip(
+            note_bank[f"{self.prefix}-no{i}"] = self.generate_note(note)
+            note_order.append(f"{self.prefix}-no{i}")
+        clip_count = math.ceil(max_quarter / 32)
+        clip_bank = {}
+        for i in range(clip_count):
+            clip_bank[f"{self.prefix}-cl{i}"] = VOXFactoryVocalClip(
+                start_quarter=32 * i,
+                offset_quarter=32 * i,
+                length=32,
                 note_bank=note_bank,
                 note_order=note_order,
-            ),
-        }
+            )
+        return clip_bank
 
     def generate_note(self, note: Note) -> VOXFactoryNote:
+        note_start_time = self.synchronizer.get_actual_secs_from_ticks(note.start_pos)
         return VOXFactoryNote(
+            time=note_start_time,
             ticks=note.start_pos,
             duration_ticks=note.length,
             midi=note.key_number,

diff --git a/libresvip/plugins/vfp/vox_factory_parser.py b/libresvip/plugins/vfp/vox_factory_parser.py
@@ -30,12 +30,11 @@ class VOXFactoryParser:
     first_bar_length: int = dataclasses.field(init=False)
 
     def parse_project(self, vox_project: VOXFactoryProject) -> Project:
-        project = Project(
+        return Project(
             song_tempo_list=self.parse_tempos(vox_project.tempo),
             time_signature_list=self.parse_time_signatures(vox_project.time_signature),
             track_list=self.parse_tracks(vox_project),
         )
-        return project
 
     def parse_tempos(self, tempo: float) -> list[SongTempo]:
         tempos = [