daw-master-dawdreamer

name	daw-master:dawdreamer
description	DawDreamer Python framework wrapper - full DAW operations: effects, mixing, time/pitch, VST plugins, multi-track composition
version	0.1.0
author	Hermes Agent
license	MIT
metadata	{"hermes":{"tags":["audio","daw","python","effects","vst","mixing","time-stretch","pitch-shift"],"related_skills":["daw-master:sox-engine","daw-master:ffmpeg-audio","daw-master:rubber-band"]}}

DawDreamer Skill

Wraps the DawDreamer Python framework (JUCE-based) for full DAW capabilities within Hermes.

Quick Status

# Check if DawDreamer is available
python -c "import dawdreamer; print(dawdreamer.__version__)"

If not installed:

# DawDreamer requires JUCE and system audio libs. Install dependencies first:
sudo apt update
sudo apt install build-essential cmake libasound2-dev libjack-jackd2-dev libcurl4-openssl-dev

# Then install via pip
pip install dawdreamer

Platform note: Pre-built wheels are only available for x86_64. On ARM (e.g., Raspberry Pi), you must build from source (pip install --no-binary :all: dawdreamer), which requires significant RAM and may fail on constrained devices. Consider cross-compiling on a more capable ARM machine or using an x86_64 host.

Overview

DawDreamer provides:

• Multi-track composition — build graphs of audio processors
• VST2/VST3 hosting — load and automate plugins
• Audio playback/recording — file-based or real-time
• Time-stretching & pitch-warping — Ableton-style warping
• Mix & render — offline rendering to WAV/MP3/FLAC

This skill wraps those capabilities in a pipeline-friendly interface.

---

Core Concept: The Pipeline

Transformations are sequences of operations applied to an audio file:

# Conceptual example
result = dawdreamer.transform(
    input="vocals.wav",
    pipeline=[
        {"op": "normalize", "target_level": -6},
        {"op": "time_stretch", "factor": 0.88, "preserve_formants": True},
        {"op": "add_reverb", "room_size": 0.4, "wet": 0.3},
        {"op": "compressor", "threshold": -20, "ratio": 4},
        {"op": "fade", "in": 0.1, "out": 2.0},
    ],
    output="vocals_processed.wav"
)

---

Operations Reference

Audio I/O

• load — Load audio file into working buffer
• save / render — Write buffer to output file
• info — Get audio properties (duration, sample_rate, channels, peak, RMS)

Time & Pitch

• time_stretch — Change tempo without affecting pitch (via Rubber Band or similar)
  • factor: tempo multiplier (0.5 = half speed, 2.0 = double)
  • preserve_formants: keep vocal character (bool)
• pitch_shift — Change pitch without affecting tempo
  • semitones: pitch shift in semitones (can be fractional)
  • preserve_formants: keep timbre
• warp — Full Ableton-style warp markers (advanced)

Level & Dynamics

• gain / volume — Apply linear gain (dB: +6, -3, etc.)
• normalize — Scale to target peak (default: -0.1 dB)
• compress — Dynamic range compression
  • threshold (dB), ratio (4:1 = 4), attack (s), release (s)
• limit — Hard limiter/ceiling

EQ & Filtering

• equalizer — Multi-band EQ
  • bands: list of {freq, gain, Q} objects
• lowpass / highpass — Simple filters
  • cutoff (Hz), order (filter steepness)
• loudness — Apply loudness normalization (EBU R128 target)

Effects

• reverb — Convolution or algorithmic reverb
  • room_size, wet, dry, decay
• delay / echo — Time-based delay
  • time (seconds), feedback, wet
• chorus / flanger — Modulation effects
  • depth, rate, mix
• distortion / overdrive — Waveshaping

Editing

• trim / crop — Extract segment
  • start (seconds), end (seconds or duration)
• fade / fade_in / fade_out — Fade amplitude
  • duration (seconds), curve (linear, exp, log)
• reverse — Reverse audio
• concat / append — Join audio files end-to-end
• overlay / mix — Blend two audio files (sample-accurate sync)

Mixing

• add_track — Load additional audio as a new track
• mix_tracks — Sum all tracks with optional per-track gain/pan
• pan / balance — Stereo positioning
• solo / mute — Track enable/disable

Analysis (Read-Only)

• analyze — Full feature extraction
  • Returns: duration, sample_rate, channels, peak_dB, RMS_dB, loudness_LUFS
  • Optional: compute_mfcc, compute_spectral_contrast
• beat_track — Estimate tempo and beats (like librosa)
• onset_detect — Find onset times
• pitch_detect — Fundamental frequency estimation

---

Tool Use: Implementation Pattern

from hermes_tools import file, execute_code, web_search

# DawDreamer skill file structure:
SKILL.md   ← this file (metadata + documentation)
dawdreamer/__init__.py   ← main implementation
dawdreamer/examples/     ← example scripts
dawdreamer/templates/    ← project templates

The skill registers with Hermes as skill:daw-master:dawdreamer (or dawdreamer if the daw-master namespace is implicit).

---

Example Scripts

Simple Time-Stretch + Fade

import dawdreamer

result = dawdreamer.transform(
    input="input.wav",
    pipeline=[
        {"op": "time_stretch", "factor": 0.88, "preserve_formants": True},
        {"op": "normalize", "target_level": -6},
        {"op": "fade_out", "duration": 1.5},
    ],
    output="slowed_faded.wav"
)
# result: {success: true, output: "slowed_faded.wav", duration: 45.2}

Add Reverb from a DSP rack (VST)

result = dawdreamer.transform(
    input="drums.wav",
    pipeline=[
        {"op": "load_vst", "path": "/usr/local/vst/ValhallaRoom.vst3"},
        {"op": "set_param", "plugin_idx": 0, "param": "RoomSize", "value": 0.75},
        {"op": "set_param", "plugin_idx": 0, "param": "Wet", "value": 0.4},
    ],
    output="drums_spacy.wav"
)

Mix Two Stems Together

result = dawdreamer.mix(
    tracks=[
        {"path": "vocals.wav", "gain_db": 0, "pan": 0.0},
        {"path": "instrumental.wav", "gain_db": -3, "pan": 0.3},
    ],
    output="full_mix.wav",
    normalize_final=True
)

Batch Process a Directory

from pathlib import Path

for wav in Path("samples/").glob("*.wav"):
    out = f"processed/{wav.stem}_proc.wav"
    dawdreamer.transform(
        input=str(wav),
        pipeline=[{"op": "normalize"}, {"op": "compress", "threshold": -20}],
        output=out
    )

---

Error Handling & Safety

• Input validation: File existence, format support, readable
• Dry-run mode: dry_run=True describes pipeline without executing
• Peak protection: Clipping prevention (clip_protection=True)
• Backup originals: backup_original=True copies input to .bak/ before processing
• Timeouts: Long renders can be bounded via timeout_seconds

---

Hermes Integration

Call from Hermes via:

Use skill daw-master:dawdreamer transform input="file.wav" pipeline=[...] output="out.wav"

Or from Python:

from hermes import skill

result = skill.call('daw-master:dawdreamer', 'transform', {
    'input': 'sample.wav',
    'pipeline': [...],
    'output': 'out.wav'
})

---

Chaining with Other Skills

DawDreamer integrates downstream/upstream with other daw-master skills:

# 1. Analyze with audio-analyzer
analysis = audio_analyzer.analyze("sample.wav")

# 2. Build pipeline based on analysis
pipeline = []
if analysis["loudness_lufs"] > -10:
    pipeline.append({"op": "gain", "amount_db": -3})  # bring down
if analysis["bpm"] < 80:
    pipeline.append({"op": "time_stretch", "factor": 1.15})  # speed up

# 3. Apply transformations
result = dawdreamer.transform("sample.wav", pipeline=pipeline, output="final.wav")

# 4. Tag with metadata
bwfmetaedit.write("final.wav", description="Processed via Hermes pipeline")

---

Notes

• DawDreamer uses JUCE's audio engine; real-time constraints apply. For long files, consider chunked rendering (chunk_size parameter).
• VST plugin paths: Set VST_PATH or VST3_PATH environment variables, or pass explicit vst_search_paths to the skill.
• Sample rate conversion: DawDreamer resamples using high-quality algorithms; no manual resample step needed unless targeting specific sample rate.
• Multi-threading: Set num_workers for batch operations to parallelize across CPU cores.