448 lines
24 KiB
Python
448 lines
24 KiB
Python
import os, logging, math
|
|
import numpy as np
|
|
import soundfile as sf
|
|
import scipy.signal as signal
|
|
from app.config import settings
|
|
from app.core.vst_engine import (
|
|
render_midi_events_to_audio,
|
|
PluginManager,
|
|
DecentSamplerManager,
|
|
HAS_PEDALBOARD,
|
|
HAS_PYFLUIDSYNTH,
|
|
)
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
UPLOAD_SF_DIR = os.path.join(settings.STORAGE_DIR, "soundfonts")
|
|
SYSTEM_SF_DIR = "/opt/daw_engine/soundfonts"
|
|
|
|
SYS_SOUNDFONTS = [
|
|
("GeneralUser_GS.sf2", "GeneralUser GS"),
|
|
("SGM_v2.01.sf2", "SGM v2.01"),
|
|
("SGM-V2.01.sf2", "SGM v2.01"),
|
|
]
|
|
|
|
def _find_sf2_path(sf_id: str) -> str:
|
|
clean_id = sf_id.replace("sf_", "") if sf_id.startswith("sf_") else sf_id
|
|
clean_lower = clean_id.lower()
|
|
sf_lower = sf_id.lower()
|
|
for base_dir in [UPLOAD_SF_DIR, SYSTEM_SF_DIR]:
|
|
if not os.path.isdir(base_dir):
|
|
continue
|
|
for fname in os.listdir(base_dir):
|
|
fbase, fext = os.path.splitext(fname)
|
|
if fext.lower() in (".sf2", ".sf3") and (fbase.lower() == clean_lower or fbase.lower() == sf_lower):
|
|
return os.path.join(base_dir, fname)
|
|
static_dir = os.path.join(settings.APP_DIR, "static", "soundfonts")
|
|
if os.path.isdir(static_dir):
|
|
for fname in os.listdir(static_dir):
|
|
fbase, fext = os.path.splitext(fname)
|
|
if fext.lower() in (".sf2", ".sf3") and (fbase.lower() == clean_lower or fbase.lower() == sf_lower):
|
|
return os.path.join(static_dir, fname)
|
|
return ""
|
|
|
|
def _find_default_sf2() -> str:
|
|
for sf_name, _ in SYS_SOUNDFONTS:
|
|
for base_dir in [SYSTEM_SF_DIR, UPLOAD_SF_DIR]:
|
|
p = os.path.join(base_dir, sf_name)
|
|
if os.path.exists(p):
|
|
return p
|
|
return ""
|
|
|
|
if HAS_PEDALBOARD:
|
|
try:
|
|
from pedalboard import Pedalboard, Gain, Chorus, Reverb
|
|
except Exception:
|
|
HAS_PEDALBOARD = False
|
|
|
|
|
|
|
|
class PythonRenderEngine:
|
|
def __init__(self, sample_rate=44100):
|
|
self.sample_rate = sample_rate
|
|
|
|
def bars_to_samples(self, bars: float, bpm: float, time_sig_num: int) -> int:
|
|
seconds_per_beat = 60.0 / max(20.0, bpm)
|
|
seconds_per_bar = seconds_per_beat * time_sig_num
|
|
return int(bars * seconds_per_bar * self.sample_rate)
|
|
|
|
def resolve_file_path(self, url_or_id: str) -> str:
|
|
if not url_or_id:
|
|
return ""
|
|
base = os.path.basename(url_or_id)
|
|
# Check uploads directory
|
|
p_uploads = os.path.join(settings.UPLOADS_DIR, base)
|
|
if os.path.exists(p_uploads):
|
|
return p_uploads
|
|
# Check processed directory
|
|
p_processed = os.path.join(settings.PROCESSED_DIR, base)
|
|
if os.path.exists(p_processed):
|
|
return p_processed
|
|
# Check general storage directory
|
|
p_storage = os.path.join(settings.STORAGE_DIR, base)
|
|
if os.path.exists(p_storage):
|
|
return p_storage
|
|
# Direct check
|
|
if os.path.exists(url_or_id):
|
|
return url_or_id
|
|
return url_or_id
|
|
|
|
def render_session_container(self, session: dict, section_store: dict, bpm: float, time_sig_num: int, total_samples: int, _cache: dict = None) -> np.ndarray:
|
|
session_buffer = np.zeros((2, total_samples), dtype=np.float32)
|
|
|
|
# Solo semantics: when any track is soloed, only soloed tracks sound.
|
|
tracks = session.get("tracks", [])
|
|
solo_ids = {t.get("id") for t in tracks if t.get("solo")}
|
|
|
|
_channel_counter = 0
|
|
|
|
for track in tracks:
|
|
if solo_ids and track.get("id") not in solo_ids:
|
|
continue
|
|
track_type = track.get("type", "AUDIO")
|
|
track_buffer = np.zeros((2, total_samples), dtype=np.float32)
|
|
|
|
# Parse synth_engine struct (Task C) — fall back to flat fields
|
|
se = track.get("synth_engine", {}) or {}
|
|
instrument_id = se.get("plugin_id") or track.get("instrument_id", "") or track.get("instrument", "")
|
|
vst_path = se.get("plugin_path") or "" # spec desktop: load VST3 từ plugin_path
|
|
instrument_source = se.get("type") or track.get("instrument_source", "soundfont")
|
|
soundfont_bank = se.get("soundfont_bank") if se.get("soundfont_bank") is not None else track.get("soundfont_bank", 0)
|
|
soundfont_program = se.get("soundfont_program") if se.get("soundfont_program") is not None else track.get("soundfont_program", 0)
|
|
soundfont_id = se.get("soundfont_id") or track.get("soundfont_id", "")
|
|
is_percussion = track.get("is_percussion", False) or (soundfont_bank == 128)
|
|
midi_channel = 9 if is_percussion else (_channel_counter % 9)
|
|
if not is_percussion:
|
|
_channel_counter += 1
|
|
|
|
for item in track.get("items", []):
|
|
start_sample = self.bars_to_samples(item["start_bar"], bpm, time_sig_num)
|
|
dur_samples = self.bars_to_samples(item["duration_bars"], bpm, time_sig_num)
|
|
offset_sample = self.bars_to_samples(item["clip_start_offset_bars"], bpm, time_sig_num)
|
|
|
|
item_type = item.get("type")
|
|
if item_type == "AUDIO_ITEM":
|
|
source_data = item.get("source_data", {})
|
|
audio_url = source_data.get("audio_file_url", "")
|
|
resolved_path = self.resolve_file_path(audio_url)
|
|
|
|
if resolved_path and os.path.exists(resolved_path):
|
|
try:
|
|
audio_data, sr = sf.read(resolved_path, dtype='float32')
|
|
if sr != self.sample_rate:
|
|
# Proper resampling: previously a silent no-op that
|
|
# played 48kHz audio at the wrong speed/pitch.
|
|
from scipy.signal import resample_poly
|
|
g = math.gcd(sr, self.sample_rate)
|
|
audio_data = resample_poly(
|
|
audio_data,
|
|
up=self.sample_rate // g,
|
|
down=sr // g,
|
|
axis=-1,
|
|
)
|
|
sr = self.sample_rate
|
|
|
|
# Handle channel mapping (Mono/Stereo)
|
|
if len(audio_data.shape) == 1:
|
|
audio_data = np.vstack([audio_data, audio_data])
|
|
else:
|
|
audio_data = audio_data.T # Shape: (channels, samples)
|
|
|
|
# Trim source offset & duration
|
|
src_len = audio_data.shape[1]
|
|
if offset_sample < src_len:
|
|
actual_dur = min(dur_samples, src_len - offset_sample)
|
|
sliced_audio = audio_data[:, offset_sample : offset_sample + actual_dur]
|
|
|
|
# Apply gain
|
|
gain_val = source_data.get("gain", 1.0)
|
|
sliced_audio = sliced_audio * gain_val
|
|
|
|
# Write to track buffer with boundaries
|
|
write_end = min(start_sample + sliced_audio.shape[1], total_samples)
|
|
actual_len = write_end - start_sample
|
|
if actual_len > 0:
|
|
track_buffer[:, start_sample:write_end] += sliced_audio[:, :actual_len]
|
|
except Exception as e:
|
|
logger.warning("[RenderEngine] Error reading audio file %s: %s", resolved_path, e)
|
|
|
|
elif item_type == "MIDI_ITEM":
|
|
source_data = item.get("source_data", {})
|
|
notes = source_data.get("notes", [])
|
|
|
|
# Convert to midi events required by vst_engine
|
|
midi_events = []
|
|
for note in notes:
|
|
note_start_bar = note["start_beat"] / time_sig_num
|
|
# Filter notes within the non-destructive visible window
|
|
offset_bar = item["clip_start_offset_bars"]
|
|
dur_bar = item["duration_bars"]
|
|
if note_start_bar >= offset_bar and note_start_bar < (offset_bar + dur_bar):
|
|
rel_bar_in_item = note_start_bar - offset_bar
|
|
target_global_bar = item["start_bar"] + rel_bar_in_item
|
|
midi_events.append({
|
|
"note": note["pitch"],
|
|
"start_beat": target_global_bar * time_sig_num,
|
|
"duration_beats": note["duration_beats"],
|
|
"velocity": int(note.get("velocity", 0.8) * 127)
|
|
})
|
|
|
|
if midi_events:
|
|
try:
|
|
plugin_mgr = PluginManager()
|
|
# Ưu tiên plugin_path (spec desktop) — fallback plugin_id theo tên
|
|
vst = plugin_mgr.load_vst(vst_path or instrument_id) if (vst_path or instrument_id) else None
|
|
|
|
if instrument_source == "pianobook":
|
|
dspreset_path = track.get("dspreset_path", "")
|
|
if dspreset_path and os.path.exists(dspreset_path) and HAS_PEDALBOARD:
|
|
from pedalboard import Pedalboard
|
|
ds_manager = DecentSamplerManager()
|
|
try:
|
|
vst = ds_manager.create_decent_sampler_instance(dspreset_path)
|
|
midi_messages = PluginManager.midi_events_to_messages(
|
|
midi_events, bpm, self.sample_rate
|
|
)
|
|
total_needed = 0
|
|
for ev in midi_events:
|
|
end_sec = (ev.get("start_beat", 0) + ev.get("duration_beats", 1)) * (60.0 / bpm)
|
|
dur_samples = int(end_sec * self.sample_rate)
|
|
if dur_samples > total_needed:
|
|
total_needed = dur_samples
|
|
total_needed = max(total_needed, 1024)
|
|
silent = np.zeros((2, total_needed), dtype=np.float32)
|
|
board = Pedalboard([vst])
|
|
synth_buffer = board(silent, sample_rate=self.sample_rate, midi_messages=midi_messages)
|
|
except Exception as e:
|
|
logger.warning(f"[RenderEngine] DecentSampler/Pianobook error: {e}")
|
|
synth_buffer = render_midi_events_to_audio(
|
|
midi_events=midi_events, sr=self.sample_rate, bpm=bpm, instrument='synth'
|
|
)
|
|
else:
|
|
synth_buffer = render_midi_events_to_audio(
|
|
midi_events=midi_events, sr=self.sample_rate, bpm=bpm, instrument='synth'
|
|
)
|
|
elif vst and HAS_PEDALBOARD:
|
|
from pedalboard import Pedalboard
|
|
midi_messages = PluginManager.midi_events_to_messages(
|
|
midi_events, bpm, self.sample_rate,
|
|
bank=soundfont_bank, program=soundfont_program
|
|
)
|
|
total_needed = 0
|
|
for ev in midi_events:
|
|
end_sec = (ev.get("start_beat", 0) + ev.get("duration_beats", 1)) * (60.0 / bpm)
|
|
dur_samples = int(end_sec * self.sample_rate)
|
|
if dur_samples > total_needed:
|
|
total_needed = dur_samples
|
|
total_needed = max(total_needed, 1024)
|
|
silent = np.zeros((2, total_needed), dtype=np.float32)
|
|
board = Pedalboard([vst])
|
|
synth_buffer = board(silent, sample_rate=self.sample_rate, midi_messages=midi_messages)
|
|
elif instrument_id and (instrument_id.startswith("sf_") or soundfont_id):
|
|
sf_path = _find_sf2_path(soundfont_id or instrument_id)
|
|
# 3-level fallback: selected SF → default SF → oscillator synth
|
|
if not sf_path or not os.path.exists(sf_path) or not HAS_PYFLUIDSYNTH:
|
|
if not sf_path or not os.path.exists(sf_path):
|
|
logger.warning(f"[RenderEngine] SoundFont not found for {soundfont_id or instrument_id}, trying default")
|
|
sf_path = _find_default_sf2() if HAS_PYFLUIDSYNTH else ""
|
|
|
|
if sf_path and os.path.exists(sf_path) and HAS_PYFLUIDSYNTH:
|
|
import fluidsynth as _fs
|
|
_settings = _fs.new_fluid_settings()
|
|
_fs.fluid_settings_setnum(_settings, b'synth.sample-rate', float(self.sample_rate))
|
|
_fl = _fs.new_fluid_synth(_settings)
|
|
_fid = _fs.fluid_synth_sfload(_fl, sf_path.encode("utf-8"), 1)
|
|
_fs.fluid_synth_program_select(_fl, midi_channel, _fid, soundfont_bank, soundfont_program)
|
|
beat_sec = 60.0 / bpm
|
|
total_sec = 0
|
|
for ev in midi_events:
|
|
end_sec = (ev.get("start_beat", 0) + ev.get("duration_beats", 1)) * beat_sec
|
|
if end_sec > total_sec:
|
|
total_sec = end_sec
|
|
sf_total_samples = int((total_sec + 1.0) * self.sample_rate)
|
|
midi_data = np.zeros((2, sf_total_samples), dtype=np.float32)
|
|
_cursor = 0
|
|
for ev in sorted(midi_events, key=lambda e: e.get("start_beat", 0)):
|
|
note = ev.get("note", 60)
|
|
velocity = ev.get("velocity", 100)
|
|
start_beat = ev.get("start_beat", 0.0)
|
|
dur_beats = ev.get("duration_beats", 1.0)
|
|
start_sec = start_beat * beat_sec
|
|
dur_sec = dur_beats * beat_sec
|
|
start_s = int(start_sec * self.sample_rate)
|
|
dur_s = int(dur_sec * self.sample_rate)
|
|
# Advance synth time by rendering silence
|
|
if start_s > _cursor:
|
|
gap = start_s - _cursor
|
|
_fs.fluid_synth_write_s16_stereo(_fl, gap)
|
|
_cursor = start_s
|
|
# Start note
|
|
_fs.fluid_synth_noteon(_fl, midi_channel, note, min(velocity, 127))
|
|
block_s16 = _fs.fluid_synth_write_s16_stereo(_fl, dur_s)
|
|
_fs.fluid_synth_noteoff(_fl, midi_channel, note)
|
|
block = block_s16.astype(np.float32).reshape(-1, 2).T / 32768.0
|
|
end_s = min(_cursor + block.shape[1], sf_total_samples)
|
|
actual = end_s - _cursor
|
|
if actual > 0 and block.shape[1] > 0:
|
|
midi_data[:, _cursor:end_s] += block[:, :actual]
|
|
_cursor = end_s
|
|
synth_buffer = midi_data
|
|
_fs.delete_fluid_synth(_fl)
|
|
else:
|
|
if not HAS_PYFLUIDSYNTH:
|
|
logger.warning("[RenderEngine] pyfluidsynth not available, falling back to oscillator synth")
|
|
synth_buffer = render_midi_events_to_audio(
|
|
midi_events=midi_events, sr=self.sample_rate, bpm=bpm, instrument='synth'
|
|
)
|
|
else:
|
|
synth_buffer = render_midi_events_to_audio(
|
|
midi_events=midi_events, sr=self.sample_rate, bpm=bpm, instrument='synth'
|
|
)
|
|
actual_len = min(synth_buffer.shape[1], total_samples)
|
|
track_buffer[:, :actual_len] += synth_buffer[:, :actual_len]
|
|
except Exception as e:
|
|
logger.warning("[RenderEngine] Error rendering MIDI: %s", e)
|
|
|
|
elif item_type == "SECTION_ITEM":
|
|
source_data = item.get("source_data", {})
|
|
sec_id = source_data.get("referenced_section_id", "")
|
|
if sec_id and sec_id in section_store:
|
|
# Render nested section recursively, cached per section id
|
|
# so repeated section instances don't re-render every time.
|
|
cache = _cache if _cache is not None else {}
|
|
if sec_id in cache:
|
|
sec_buffer = cache[sec_id]
|
|
else:
|
|
sec_buffer = self.render_session_container(
|
|
session=section_store[sec_id],
|
|
section_store=section_store,
|
|
bpm=bpm,
|
|
time_sig_num=time_sig_num,
|
|
total_samples=total_samples,
|
|
_cache=cache,
|
|
)
|
|
cache[sec_id] = sec_buffer
|
|
|
|
# Apply non-destructive crop/slicing on section buffer
|
|
if offset_sample < total_samples:
|
|
actual_dur = min(dur_samples, total_samples - offset_sample)
|
|
sliced_sec = sec_buffer[:, offset_sample : offset_sample + actual_dur]
|
|
|
|
# Write to track buffer
|
|
write_end = min(start_sample + sliced_sec.shape[1], total_samples)
|
|
actual_len = write_end - start_sample
|
|
if actual_len > 0:
|
|
track_buffer[:, start_sample:write_end] += sliced_sec[:, :actual_len]
|
|
|
|
# Apply Track Gain (via Pedalboard or fallback)
|
|
vol_db = track.get("volume_db", 0.0)
|
|
pan = track.get("pan", 0.0)
|
|
mute = track.get("mute", False)
|
|
|
|
if mute:
|
|
continue
|
|
|
|
# Apply Track FX (Chorus or Reverb)
|
|
fx_type = track.get("fx_type")
|
|
if fx_type == "chorus":
|
|
if HAS_PEDALBOARD:
|
|
try:
|
|
board = Pedalboard([Chorus(rate_hz=1.5, depth=0.25)])
|
|
track_buffer = board(track_buffer, sample_rate=self.sample_rate)
|
|
except Exception as e:
|
|
logger.warning("[RenderEngine] Pedalboard Chorus failed: %s", e)
|
|
else:
|
|
# Fallback chorus using simple LFO delay modulation in scipy/numpy
|
|
try:
|
|
# 1.5 Hz sine LFO, modulating delay time between 15ms and 25ms (average 20ms)
|
|
lfo = 0.020 + 0.005 * np.sin(2 * np.pi * 1.5 * np.arange(total_samples) / self.sample_rate)
|
|
dry = track_buffer * 0.6
|
|
wet = np.zeros_like(track_buffer)
|
|
for ch in range(2):
|
|
indices = np.arange(total_samples) - (lfo * self.sample_rate)
|
|
indices = np.clip(indices, 0, total_samples - 1).astype(np.int32)
|
|
wet[ch, :] = track_buffer[ch, indices]
|
|
track_buffer = dry + wet * 0.5
|
|
except Exception as e:
|
|
logger.warning("[RenderEngine] Fallback Chorus failed: %s", e)
|
|
elif fx_type == "reverb":
|
|
if HAS_PEDALBOARD:
|
|
try:
|
|
board = Pedalboard([Reverb(room_size=0.5, wet_level=0.4, dry_level=0.6)])
|
|
track_buffer = board(track_buffer, sample_rate=self.sample_rate)
|
|
except Exception as e:
|
|
logger.warning("[RenderEngine] Pedalboard Reverb failed: %s", e)
|
|
else:
|
|
# Fallback reverb using exponentially decaying noise room impulse response
|
|
try:
|
|
# Generate impulse response (decaying noise)
|
|
len_ir = int(self.sample_rate * 2.0)
|
|
t_ir = np.arange(len_ir) / self.sample_rate
|
|
decay = np.exp(-t_ir / 0.5)
|
|
ir_l = (np.random.rand(len_ir) * 2 - 1) * decay
|
|
ir_r = (np.random.rand(len_ir) * 2 - 1) * decay
|
|
|
|
dry = track_buffer * 0.6
|
|
wet = np.zeros_like(track_buffer)
|
|
for ch in range(2):
|
|
ir = ir_l if ch == 0 else ir_r
|
|
# Convolve
|
|
conv = signal.convolve(track_buffer[ch, :], ir, mode='full')[:total_samples]
|
|
wet[ch, :] = conv
|
|
track_buffer = dry + wet * 0.4
|
|
except Exception as e:
|
|
logger.warning("[RenderEngine] Fallback Reverb failed: %s", e)
|
|
|
|
# Process track volume
|
|
if HAS_PEDALBOARD:
|
|
try:
|
|
board = Pedalboard([Gain(gain_db=vol_db)])
|
|
processed_track = board(track_buffer, sample_rate=self.sample_rate)
|
|
except Exception:
|
|
gain_linear = 10 ** (vol_db / 20.0)
|
|
processed_track = track_buffer * gain_linear
|
|
else:
|
|
gain_linear = 10 ** (vol_db / 20.0)
|
|
processed_track = track_buffer * gain_linear
|
|
|
|
# Apply Track Pan
|
|
if pan != 0.0:
|
|
# Constant power panning
|
|
theta = ((np.clip(pan, -1.0, 1.0) + 1.0) / 2.0) * (np.pi / 2.0)
|
|
processed_track[0, :] *= np.cos(theta)
|
|
processed_track[1, :] *= np.sin(theta)
|
|
|
|
# Mix track to session
|
|
session_buffer += processed_track
|
|
|
|
return session_buffer
|
|
|
|
def render_project(self, project_json: dict, output_filepath: str):
|
|
bpm = project_json["metadata"]["bpm"]
|
|
time_sig_num = project_json["metadata"].get("time_signature_numerator", 4)
|
|
main_session = project_json["main_session"]
|
|
section_store = project_json.get("section_store", {})
|
|
|
|
# Compute total project samples
|
|
total_bars = main_session.get("length_bars", 16.0)
|
|
total_samples = self.bars_to_samples(total_bars, bpm, time_sig_num)
|
|
|
|
# Render main session
|
|
master_buffer = self.render_session_container(
|
|
session=main_session,
|
|
section_store=section_store,
|
|
bpm=bpm,
|
|
time_sig_num=time_sig_num,
|
|
total_samples=total_samples,
|
|
_cache={},
|
|
)
|
|
|
|
# Normalization to prevent clipping
|
|
max_peak = np.max(np.abs(master_buffer))
|
|
if max_peak > 1.0:
|
|
master_buffer /= max_peak
|
|
|
|
# Write final output file
|
|
sf.write(output_filepath, master_buffer.T, self.sample_rate)
|
|
return output_filepath
|