Upload 3 files

RC_Vocal_Textures_Full_version.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:20dedc588c83d8d65eb6375f6890a339f19f9bc6bb785a6230adf2bfe7c7a117
+size 4854122382

RC_Vocal_Textures_Small_version.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ebd6a78509508702d57bd75f8467caa85caf0efe9e265e16968d4ee7e16249e6
+size 2427165507

model_config.json

@@ -0,0 +1,125 @@
+{
+    "model_type": "diffusion_cond",
+    "sample_size": 882000,
+    "sample_rate": 44100,
+    "audio_channels": 2,
+    "model": {
+        "pretransform": {
+            "type": "autoencoder",
+            "iterate_batch": true,
+            "config": {
+                "encoder": {
+                    "type": "oobleck",
+                    "requires_grad": false,
+                    "config": {
+                        "in_channels": 2,
+                        "channels": 128,
+                        "c_mults": [1, 2, 4, 8, 16],
+                        "strides": [2, 4, 4, 8, 8],
+                        "latent_dim": 128,
+                        "use_snake": true
+                    }
+                },
+                "decoder": {
+                    "type": "oobleck",
+                    "config": {
+                        "out_channels": 2,
+                        "channels": 128,
+                        "c_mults": [1, 2, 4, 8, 16],
+                        "strides": [2, 4, 4, 8, 8],
+                        "latent_dim": 64,
+                        "use_snake": true,
+                        "final_tanh": false
+                    }
+                },
+                "bottleneck": {
+                    "type": "vae"
+                },
+                "latent_dim": 64,
+                "downsampling_ratio": 2048,
+                "io_channels": 2
+            }
+        },
+        "conditioning": {
+            "configs": [
+                {
+                    "id": "prompt",
+                    "type": "t5",
+                    "config": {
+                        "t5_model_name": "t5-base",
+                        "max_length": 128
+                    }
+                },
+                {
+                    "id": "seconds_start",
+                    "type": "number",
+                    "config": {
+                        "min_val": 0,
+                        "max_val": 512
+                    }
+                },
+                {
+                    "id": "seconds_total",
+                    "type": "number",
+                    "config": {
+                        "min_val": 0,
+                        "max_val": 512
+                    }
+                }
+            ],
+            "cond_dim": 768
+        },
+        "diffusion": {
+            "cross_attention_cond_ids": ["prompt", "seconds_start", "seconds_total"],
+            "global_cond_ids": ["seconds_start", "seconds_total"],
+            "type": "dit",
+            "config": {
+                "io_channels": 64,
+                "embed_dim": 1536,
+                "depth": 24,
+                "num_heads": 24,
+                "cond_token_dim": 768,
+                "global_cond_dim": 1536,
+                "project_cond_tokens": false,
+                "transformer_type": "continuous_transformer"
+            }
+        },
+        "io_channels": 64
+    },
+    "training": {
+        "use_ema": true,
+        "log_loss_info": false,
+        "optimizer_configs": {
+            "diffusion": {
+                "optimizer": {
+                    "type": "AdamW",
+                    "config": {
+                        "lr": 5e-5,
+                        "betas": [0.9, 0.999],
+                        "weight_decay": 1e-3
+                    }
+                },
+                "scheduler": {
+                    "type": "InverseLR",
+                    "config": {
+                        "inv_gamma": 1000000,
+                        "power": 0.5,
+                        "warmup": 0.99
+                    }
+                }
+            }
+        },
+        "demo": {
+            "demo_every": 150,
+            "demo_steps": 250,
+            "num_demos": 4,
+            "demo_cond": [
+                {"prompt": "Ensemble Vocal Texture, chord progression, G# major, 128BPM, 8 bars", "seconds_start": 0, "seconds_total": 15},
+                {"prompt": "Female Vocal Texture, chord progression, D# minor, 110BPM, 8 bars", "seconds_start": 0, "seconds_total": 17},
+                {"prompt": "Male Vocal Texture, chord progression, C# major, 140BPM, 4 bars", "seconds_start": 0, "seconds_total": 7},
+                {"prompt": "Ensemble Vocal Texture, chord progression, G# major, 128BPM, 8 bars", "seconds_start": 0, "seconds_total": 15}
+            ],
+            "demo_cfg_scales": [4, 7]
+        }
+    }
+}