{ "cells": [ { "cell_type": "code", "execution_count": 1, "metadata": {}, "outputs": [], "source": [ "from suno_utils.utils.s3 import read_from_s3\n", "from suno_utils.audio import Audio\n", "import numpy as np" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "base_path = \"s3://suno-data/christian/outputs/mountain-whisper-eclipse/VveQ3alEmXE/VveQ3alEmXE.mp3\"\n", "item_id = \"VveQ3alEmXE\" \n", "\n", "generated_mp3 = f\"{base_path}/{item_id}.mp3\"\n", "audio = Audio.from_s3(generated_mp3)\n", "audio.play()\n", "\n", "metadata = read_from_s3(f\"{base_path}/{item_id}_metadata.json\")\n", "print(metadata)\n", "\n", "gen_semantic = read_from_s3(f\"{base_path}/{item_id}_generated_semantic.npz\", read_f=np.load)[\"semantic_codes\"]\n", "print(gen_semantic.shape)\n", "\n", "original_semantic = read_from_s3(f\"{base_path}/original_semantic.npz\", read_f=np.load)[\"semantic_codes\"]\n", "print(original_semantic)\n", "\n", "\n" ] }, { "cell_type": "code", "execution_count": 6, "metadata": {}, "outputs": [], "source": [ "\n", "import torch\n", "import math\n", "\n", "def append_zero(x):\n", " return torch.cat([x, x.new_zeros([1])])\n", "\n", "def sigma_to_t(sigma):\n", " return sigma.atan() / math.pi * 2\n", "\n", "def t_to_sigma(t):\n", " return (t * math.pi / 2).tan()\n", "\n", "def get_sigmas_polyexponential(n, sigma_min, sigma_max, rho=1.0, device=\"cpu\"):\n", " \"\"\"Constructs an polynomial in log sigma noise schedule.\"\"\"\n", " ramp = torch.linspace(1, 0, n, device=device) ** rho\n", " sigmas = torch.exp(ramp * (math.log(sigma_max) - math.log(sigma_min)) + math.log(sigma_min))\n", " return append_zero(sigmas)" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "\n", "steps = 3\n", "sigma_min = 0.5\n", "sigma_max = 100\n", "rho = 1.0\n", "\n", "sigmas = get_sigmas_polyexponential(steps, sigma_min, sigma_max, rho, device=\"cuda\")\n", "print(sigmas)\n", "\n", "t = sigma_to_t(sigmas)\n", "print(t)\n" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "t = torch.tensor([0.999, 0.749, 0.499, 0.249])\n", "sigmas = t_to_sigma(t)\n", "print(sigmas)\n" ] }, { "cell_type": "code", "execution_count": 2, "metadata": {}, "outputs": [], "source": [ "import os \n", "\n", "def setup_datasets(data_dir: str, val_size: int = 100):\n", " item_dirs = [\n", " d for d in os.listdir(data_dir) if os.path.isdir(os.path.join(data_dir, d))\n", " ]\n", "\n", " # valid item dirs are those that contain both original and generated semantic files\n", " item_ids = [\n", " os.path.basename(d)\n", " for d in item_dirs\n", " if os.path.exists(os.path.join(data_dir, d, f\"{d}_original_semantic.npz\"))\n", " and os.path.exists(os.path.join(data_dir, d, f\"{d}_generated_semantic.npz\"))\n", " ]\n", "\n", " # sort the item_ids by name\n", " item_ids.sort()\n", "\n", " # split into train and val\n", " train_item_ids = item_ids[:-val_size]\n", " val_item_ids = item_ids[-val_size:]\n", "\n", " print(f\"Found {len(item_ids)} item ids in {data_dir}\")\n", " print(f\"Splitting into {len(train_item_ids)} train and {len(val_item_ids)} val\")\n", "\n", " return train_item_ids, val_item_ids\n" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "# test loading everything into ram for now\n", "\n", "data_dir = \"/mnt/localdisk/christian\"\n", "\n", "train_item_ids, val_item_ids = setup_datasets(data_dir, val_size=100)\n" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "# load train set \n", "import torch\n", "from tqdm import tqdm\n", "\n", "examples = []\n", "for item_id in tqdm(train_item_ids):\n", " # read from local disk\n", " original_semantic = np.load(f\"{data_dir}/{item_id}/{item_id}_original_semantic.npz\")[\"semantic_codes\"]\n", " generated_semantic = np.load(f\"{data_dir}/{item_id}/{item_id}_generated_semantic.npz\")[\"semantic_codes\"]\n", " # convert to torch\n", " original_semantic = torch.from_numpy(original_semantic).long()\n", " generated_semantic = torch.from_numpy(generated_semantic).long()\n", " examples.append((original_semantic, generated_semantic))\n", "\n", "\n", "print(len(examples))\n" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "from suno_utils.utils.s3 import list_s3_dir\n", "from suno_utils.utils.text import read_jsonl\n", "\n", "output_str = \"mountain-whisper-eclipse\"\n", "\n", "work_items = read_jsonl(\n", " \"/home/christian/code/christian/metadata/discogs_subset_sampled_metas.jsonl\"\n", ")\n", "print(len(work_items))\n", "\n", "# first check for existing ids in the output path\n", "existing_ids = list_s3_dir(\n", " f\"s3://suno-data/christian/outputs/{output_str}/\",\n", ")\n", "\n", "existing_ids = [os.path.dirname(f[0]).split(\"/\")[-1] for f in existing_ids]\n", "existing_ids = list(set(existing_ids))\n", "print(len(existing_ids))\n", "print(existing_ids[:3])\n", "\n", "# now remove these from the work items\n", "work_items = [f for f in work_items if f[\"id\"] not in existing_ids]\n", "print(len(work_items))\n" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [] } ], "metadata": { "kernelspec": { "display_name": "suno_diff", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.12.9" } }, "nbformat": 4, "nbformat_minor": 2 }