facebookresearch--audiocraft
Alexandre Défossez (autodoc)
573 行
31 KiB
HTML
573 行
31 KiB
HTML
<!doctype html>
|
||
<html lang="en">
|
||
<head>
|
||
<meta charset="utf-8">
|
||
<meta name="viewport" content="width=device-width, initial-scale=1, minimum-scale=1">
|
||
<meta name="generator" content="pdoc3 0.11.5">
|
||
<title>audiocraft.models.builders API documentation</title>
|
||
<meta name="description" content="All the functions to build the relevant models and modules
|
||
from the Hydra config.">
|
||
<link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/10up-sanitize.css/13.0.0/sanitize.min.css" integrity="sha512-y1dtMcuvtTMJc1yPgEqF0ZjQbhnc/bFhyvIyVNb9Zk5mIGtqVaAB1Ttl28su8AvFMOY0EwRbAe+HCLqj6W7/KA==" crossorigin>
|
||
<link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/10up-sanitize.css/13.0.0/typography.min.css" integrity="sha512-Y1DYSb995BAfxobCkKepB1BqJJTPrOp3zPL74AWFugHHmmdcvO+C48WLrUOlhGMc0QG7AE3f7gmvvcrmX2fDoA==" crossorigin>
|
||
<link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/highlight.js/11.9.0/styles/default.min.css" crossorigin>
|
||
<style>:root{--highlight-color:#fe9}.flex{display:flex !important}body{line-height:1.5em}#content{padding:20px}#sidebar{padding:1.5em;overflow:hidden}#sidebar > *:last-child{margin-bottom:2cm}.http-server-breadcrumbs{font-size:130%;margin:0 0 15px 0}#footer{font-size:.75em;padding:5px 30px;border-top:1px solid #ddd;text-align:right}#footer p{margin:0 0 0 1em;display:inline-block}#footer p:last-child{margin-right:30px}h1,h2,h3,h4,h5{font-weight:300}h1{font-size:2.5em;line-height:1.1em}h2{font-size:1.75em;margin:2em 0 .50em 0}h3{font-size:1.4em;margin:1.6em 0 .7em 0}h4{margin:0;font-size:105%}h1:target,h2:target,h3:target,h4:target,h5:target,h6:target{background:var(--highlight-color);padding:.2em 0}a{color:#058;text-decoration:none;transition:color .2s ease-in-out}a:visited{color:#503}a:hover{color:#b62}.title code{font-weight:bold}h2[id^="header-"]{margin-top:2em}.ident{color:#900;font-weight:bold}pre code{font-size:.8em;line-height:1.4em;padding:1em;display:block}code{background:#f3f3f3;font-family:"DejaVu Sans Mono",monospace;padding:1px 4px;overflow-wrap:break-word}h1 code{background:transparent}pre{border-top:1px solid #ccc;border-bottom:1px solid #ccc;margin:1em 0}#http-server-module-list{display:flex;flex-flow:column}#http-server-module-list div{display:flex}#http-server-module-list dt{min-width:10%}#http-server-module-list p{margin-top:0}.toc ul,#index{list-style-type:none;margin:0;padding:0}#index code{background:transparent}#index h3{border-bottom:1px solid #ddd}#index ul{padding:0}#index h4{margin-top:.6em;font-weight:bold}@media (min-width:200ex){#index .two-column{column-count:2}}@media (min-width:300ex){#index .two-column{column-count:3}}dl{margin-bottom:2em}dl dl:last-child{margin-bottom:4em}dd{margin:0 0 1em 3em}#header-classes + dl > dd{margin-bottom:3em}dd dd{margin-left:2em}dd p{margin:10px 0}.name{background:#eee;font-size:.85em;padding:5px 10px;display:inline-block;min-width:40%}.name:hover{background:#e0e0e0}dt:target .name{background:var(--highlight-color)}.name > span:first-child{white-space:nowrap}.name.class > span:nth-child(2){margin-left:.4em}.inherited{color:#999;border-left:5px solid #eee;padding-left:1em}.inheritance em{font-style:normal;font-weight:bold}.desc h2{font-weight:400;font-size:1.25em}.desc h3{font-size:1em}.desc dt code{background:inherit}.source > summary,.git-link-div{color:#666;text-align:right;font-weight:400;font-size:.8em;text-transform:uppercase}.source summary > *{white-space:nowrap;cursor:pointer}.git-link{color:inherit;margin-left:1em}.source pre{max-height:500px;overflow:auto;margin:0}.source pre code{font-size:12px;overflow:visible;min-width:max-content}.hlist{list-style:none}.hlist li{display:inline}.hlist li:after{content:',\2002'}.hlist li:last-child:after{content:none}.hlist .hlist{display:inline;padding-left:1em}img{max-width:100%}td{padding:0 .5em}.admonition{padding:.1em 1em;margin:1em 0}.admonition-title{font-weight:bold}.admonition.note,.admonition.info,.admonition.important{background:#aef}.admonition.todo,.admonition.versionadded,.admonition.tip,.admonition.hint{background:#dfd}.admonition.warning,.admonition.versionchanged,.admonition.deprecated{background:#fd4}.admonition.error,.admonition.danger,.admonition.caution{background:lightpink}</style>
|
||
<style media="screen and (min-width: 700px)">@media screen and (min-width:700px){#sidebar{width:30%;height:100vh;overflow:auto;position:sticky;top:0}#content{width:70%;max-width:100ch;padding:3em 4em;border-left:1px solid #ddd}pre code{font-size:1em}.name{font-size:1em}main{display:flex;flex-direction:row-reverse;justify-content:flex-end}.toc ul ul,#index ul ul{padding-left:1em}.toc > ul > li{margin-top:.5em}}</style>
|
||
<style media="print">@media print{#sidebar h1{page-break-before:always}.source{display:none}}@media print{*{background:transparent !important;color:#000 !important;box-shadow:none !important;text-shadow:none !important}a[href]:after{content:" (" attr(href) ")";font-size:90%}a[href][title]:after{content:none}abbr[title]:after{content:" (" attr(title) ")"}.ir a:after,a[href^="javascript:"]:after,a[href^="#"]:after{content:""}pre,blockquote{border:1px solid #999;page-break-inside:avoid}thead{display:table-header-group}tr,img{page-break-inside:avoid}img{max-width:100% !important}@page{margin:0.5cm}p,h2,h3{orphans:3;widows:3}h1,h2,h3,h4,h5,h6{page-break-after:avoid}}</style>
|
||
<script defer src="https://cdnjs.cloudflare.com/ajax/libs/highlight.js/11.9.0/highlight.min.js" integrity="sha512-D9gUyxqja7hBtkWpPWGt9wfbfaMGVt9gnyCvYa+jojwwPHLCzUm5i8rpk7vD7wNee9bA35eYIjobYPaQuKS1MQ==" crossorigin></script>
|
||
<script>window.addEventListener('DOMContentLoaded', () => {
|
||
hljs.configure({languages: ['bash', 'css', 'diff', 'graphql', 'ini', 'javascript', 'json', 'plaintext', 'python', 'python-repl', 'rust', 'shell', 'sql', 'typescript', 'xml', 'yaml']});
|
||
hljs.highlightAll();
|
||
/* Collapse source docstrings */
|
||
setTimeout(() => {
|
||
[...document.querySelectorAll('.hljs.language-python > .hljs-string')]
|
||
.filter(el => el.innerHTML.length > 200 && ['"""', "'''"].includes(el.innerHTML.substring(0, 3)))
|
||
.forEach(el => {
|
||
let d = document.createElement('details');
|
||
d.classList.add('hljs-string');
|
||
d.innerHTML = '<summary>"""</summary>' + el.innerHTML.substring(3);
|
||
el.replaceWith(d);
|
||
});
|
||
}, 100);
|
||
})</script>
|
||
</head>
|
||
<body>
|
||
<main>
|
||
<article id="content">
|
||
<header>
|
||
<h1 class="title">Module <code>audiocraft.models.builders</code></h1>
|
||
</header>
|
||
<section id="section-intro">
|
||
<p>All the functions to build the relevant models and modules
|
||
from the Hydra config.</p>
|
||
</section>
|
||
<section>
|
||
</section>
|
||
<section>
|
||
</section>
|
||
<section>
|
||
<h2 class="section-title" id="header-functions">Functions</h2>
|
||
<dl>
|
||
<dt id="audiocraft.models.builders.get_codebooks_pattern_provider"><code class="name flex">
|
||
<span>def <span class="ident">get_codebooks_pattern_provider</span></span>(<span>n_q: int, cfg: omegaconf.dictconfig.DictConfig) ‑> <a title="audiocraft.modules.codebooks_patterns.CodebooksPatternProvider" href="../modules/codebooks_patterns.html#audiocraft.modules.codebooks_patterns.CodebooksPatternProvider">CodebooksPatternProvider</a></span>
|
||
</code></dt>
|
||
<dd>
|
||
<details class="source">
|
||
<summary>
|
||
<span>Expand source code</span>
|
||
</summary>
|
||
<pre><code class="python">def get_codebooks_pattern_provider(
|
||
n_q: int, cfg: omegaconf.DictConfig
|
||
) -> CodebooksPatternProvider:
|
||
"""Instantiate a codebooks pattern provider object."""
|
||
pattern_providers = {
|
||
"parallel": ParallelPatternProvider,
|
||
"delay": DelayedPatternProvider,
|
||
"unroll": UnrolledPatternProvider,
|
||
"coarse_first": CoarseFirstPattern,
|
||
"musiclm": MusicLMPattern,
|
||
}
|
||
name = cfg.modeling
|
||
kwargs = dict_from_config(cfg.get(name)) if hasattr(cfg, name) else {}
|
||
klass = pattern_providers[name]
|
||
return klass(n_q, **kwargs)</code></pre>
|
||
</details>
|
||
<div class="desc"><p>Instantiate a codebooks pattern provider object.</p></div>
|
||
</dd>
|
||
<dt id="audiocraft.models.builders.get_compression_model"><code class="name flex">
|
||
<span>def <span class="ident">get_compression_model</span></span>(<span>cfg: omegaconf.dictconfig.DictConfig) ‑> <a title="audiocraft.models.encodec.CompressionModel" href="encodec.html#audiocraft.models.encodec.CompressionModel">CompressionModel</a></span>
|
||
</code></dt>
|
||
<dd>
|
||
<details class="source">
|
||
<summary>
|
||
<span>Expand source code</span>
|
||
</summary>
|
||
<pre><code class="python">def get_compression_model(cfg: omegaconf.DictConfig) -> CompressionModel:
|
||
"""Instantiate a compression model."""
|
||
if cfg.compression_model == "encodec":
|
||
kwargs = dict_from_config(getattr(cfg, "encodec"))
|
||
encoder_name = kwargs.pop("autoencoder")
|
||
quantizer_name = kwargs.pop("quantizer")
|
||
encoder, decoder = get_encodec_autoencoder(encoder_name, cfg)
|
||
quantizer = get_quantizer(quantizer_name, cfg, encoder.dimension)
|
||
frame_rate = kwargs["sample_rate"] // encoder.hop_length
|
||
renormalize = kwargs.pop("renormalize", False)
|
||
# deprecated params
|
||
kwargs.pop("renorm", None)
|
||
return EncodecModel(
|
||
encoder,
|
||
decoder,
|
||
quantizer,
|
||
frame_rate=frame_rate,
|
||
renormalize=renormalize,
|
||
**kwargs,
|
||
).to(cfg.device)
|
||
else:
|
||
raise KeyError(f"Unexpected compression model {cfg.compression_model}")</code></pre>
|
||
</details>
|
||
<div class="desc"><p>Instantiate a compression model.</p></div>
|
||
</dd>
|
||
<dt id="audiocraft.models.builders.get_condition_fuser"><code class="name flex">
|
||
<span>def <span class="ident">get_condition_fuser</span></span>(<span>cfg: omegaconf.dictconfig.DictConfig) ‑> <a title="audiocraft.modules.conditioners.ConditionFuser" href="../modules/conditioners.html#audiocraft.modules.conditioners.ConditionFuser">ConditionFuser</a></span>
|
||
</code></dt>
|
||
<dd>
|
||
<details class="source">
|
||
<summary>
|
||
<span>Expand source code</span>
|
||
</summary>
|
||
<pre><code class="python">def get_condition_fuser(cfg: omegaconf.DictConfig) -> ConditionFuser:
|
||
"""Instantiate a condition fuser object."""
|
||
fuser_cfg = getattr(cfg, "fuser")
|
||
fuser_methods = ["sum", "cross", "prepend", "ignore", "input_interpolate"]
|
||
fuse2cond = {k: fuser_cfg[k] for k in fuser_methods if k in fuser_cfg}
|
||
kwargs = {k: v for k, v in fuser_cfg.items() if k not in fuser_methods}
|
||
fuser = ConditionFuser(fuse2cond=fuse2cond, **kwargs)
|
||
return fuser</code></pre>
|
||
</details>
|
||
<div class="desc"><p>Instantiate a condition fuser object.</p></div>
|
||
</dd>
|
||
<dt id="audiocraft.models.builders.get_conditioner_provider"><code class="name flex">
|
||
<span>def <span class="ident">get_conditioner_provider</span></span>(<span>output_dim: int, cfg: omegaconf.dictconfig.DictConfig) ‑> <a title="audiocraft.modules.conditioners.ConditioningProvider" href="../modules/conditioners.html#audiocraft.modules.conditioners.ConditioningProvider">ConditioningProvider</a></span>
|
||
</code></dt>
|
||
<dd>
|
||
<details class="source">
|
||
<summary>
|
||
<span>Expand source code</span>
|
||
</summary>
|
||
<pre><code class="python">def get_conditioner_provider(
|
||
output_dim: int, cfg: omegaconf.DictConfig
|
||
) -> ConditioningProvider:
|
||
"""Instantiate a conditioning model."""
|
||
device = cfg.device
|
||
duration = cfg.dataset.segment_duration
|
||
cfg = getattr(cfg, "conditioners")
|
||
dict_cfg = {} if cfg is None else dict_from_config(cfg)
|
||
conditioners: tp.Dict[str, BaseConditioner] = {}
|
||
condition_provider_args = dict_cfg.pop("args", {})
|
||
condition_provider_args.pop("merge_text_conditions_p", None)
|
||
condition_provider_args.pop("drop_desc_p", None)
|
||
|
||
for cond, cond_cfg in dict_cfg.items():
|
||
model_type = cond_cfg["model"]
|
||
model_args = cond_cfg[model_type]
|
||
if model_type == "t5":
|
||
conditioners[str(cond)] = T5Conditioner(
|
||
output_dim=output_dim, device=device, **model_args
|
||
)
|
||
elif model_type == "lut":
|
||
conditioners[str(cond)] = LUTConditioner(
|
||
output_dim=output_dim, **model_args
|
||
)
|
||
elif model_type == "chroma_stem":
|
||
conditioners[str(cond)] = ChromaStemConditioner(
|
||
output_dim=output_dim, duration=duration, device=device, **model_args
|
||
)
|
||
elif model_type in {"chords_emb", "drum_latents", "melody"}:
|
||
conditioners_classes = {"chords_emb": ChordsEmbConditioner,
|
||
"drum_latents": DrumsConditioner,
|
||
"melody": MelodyConditioner}
|
||
conditioner_class = conditioners_classes[model_type]
|
||
conditioners[str(cond)] = conditioner_class(device=device, **model_args)
|
||
elif model_type == "clap":
|
||
conditioners[str(cond)] = CLAPEmbeddingConditioner(
|
||
output_dim=output_dim, device=device, **model_args
|
||
)
|
||
elif model_type == 'style':
|
||
conditioners[str(cond)] = StyleConditioner(
|
||
output_dim=output_dim,
|
||
device=device,
|
||
**model_args
|
||
)
|
||
else:
|
||
raise ValueError(f"Unrecognized conditioning model: {model_type}")
|
||
conditioner = ConditioningProvider(
|
||
conditioners, device=device, **condition_provider_args
|
||
)
|
||
return conditioner</code></pre>
|
||
</details>
|
||
<div class="desc"><p>Instantiate a conditioning model.</p></div>
|
||
</dd>
|
||
<dt id="audiocraft.models.builders.get_debug_compression_model"><code class="name flex">
|
||
<span>def <span class="ident">get_debug_compression_model</span></span>(<span>device='cpu', sample_rate: int = 32000)</span>
|
||
</code></dt>
|
||
<dd>
|
||
<details class="source">
|
||
<summary>
|
||
<span>Expand source code</span>
|
||
</summary>
|
||
<pre><code class="python">def get_debug_compression_model(device="cpu", sample_rate: int = 32000):
|
||
"""Instantiate a debug compression model to be used for unit tests."""
|
||
assert sample_rate in [
|
||
16000,
|
||
32000,
|
||
], "unsupported sample rate for debug compression model"
|
||
model_ratios = {
|
||
16000: [10, 8, 8], # 25 Hz at 16kHz
|
||
32000: [10, 8, 16], # 25 Hz at 32kHz
|
||
}
|
||
ratios: tp.List[int] = model_ratios[sample_rate]
|
||
frame_rate = 25
|
||
seanet_kwargs: dict = {
|
||
"n_filters": 4,
|
||
"n_residual_layers": 1,
|
||
"dimension": 32,
|
||
"ratios": ratios,
|
||
}
|
||
encoder = audiocraft.modules.SEANetEncoder(**seanet_kwargs)
|
||
decoder = audiocraft.modules.SEANetDecoder(**seanet_kwargs)
|
||
quantizer = qt.ResidualVectorQuantizer(dimension=32, bins=400, n_q=4)
|
||
init_x = torch.randn(8, 32, 128)
|
||
quantizer(init_x, 1) # initialize kmeans etc.
|
||
compression_model = EncodecModel(
|
||
encoder,
|
||
decoder,
|
||
quantizer,
|
||
frame_rate=frame_rate,
|
||
sample_rate=sample_rate,
|
||
channels=1,
|
||
).to(device)
|
||
return compression_model.eval()</code></pre>
|
||
</details>
|
||
<div class="desc"><p>Instantiate a debug compression model to be used for unit tests.</p></div>
|
||
</dd>
|
||
<dt id="audiocraft.models.builders.get_debug_lm_model"><code class="name flex">
|
||
<span>def <span class="ident">get_debug_lm_model</span></span>(<span>device='cpu')</span>
|
||
</code></dt>
|
||
<dd>
|
||
<details class="source">
|
||
<summary>
|
||
<span>Expand source code</span>
|
||
</summary>
|
||
<pre><code class="python">def get_debug_lm_model(device="cpu"):
|
||
"""Instantiate a debug LM to be used for unit tests."""
|
||
pattern = DelayedPatternProvider(n_q=4)
|
||
dim = 16
|
||
providers = {
|
||
"description": LUTConditioner(
|
||
n_bins=128, dim=dim, output_dim=dim, tokenizer="whitespace"
|
||
),
|
||
}
|
||
condition_provider = ConditioningProvider(providers)
|
||
fuser = ConditionFuser(
|
||
{"cross": ["description"], "prepend": [], "sum": [], "input_interpolate": []}
|
||
)
|
||
lm = LMModel(
|
||
pattern,
|
||
condition_provider,
|
||
fuser,
|
||
n_q=4,
|
||
card=400,
|
||
dim=dim,
|
||
num_heads=4,
|
||
custom=True,
|
||
num_layers=2,
|
||
cross_attention=True,
|
||
causal=True,
|
||
)
|
||
return lm.to(device).eval()</code></pre>
|
||
</details>
|
||
<div class="desc"><p>Instantiate a debug LM to be used for unit tests.</p></div>
|
||
</dd>
|
||
<dt id="audiocraft.models.builders.get_diffusion_model"><code class="name flex">
|
||
<span>def <span class="ident">get_diffusion_model</span></span>(<span>cfg: omegaconf.dictconfig.DictConfig)</span>
|
||
</code></dt>
|
||
<dd>
|
||
<details class="source">
|
||
<summary>
|
||
<span>Expand source code</span>
|
||
</summary>
|
||
<pre><code class="python">def get_diffusion_model(cfg: omegaconf.DictConfig):
|
||
# TODO Find a way to infer the channels from dset
|
||
channels = cfg.channels
|
||
num_steps = cfg.schedule.num_steps
|
||
return DiffusionUnet(chin=channels, num_steps=num_steps, **cfg.diffusion_unet)</code></pre>
|
||
</details>
|
||
<div class="desc"></div>
|
||
</dd>
|
||
<dt id="audiocraft.models.builders.get_encodec_autoencoder"><code class="name flex">
|
||
<span>def <span class="ident">get_encodec_autoencoder</span></span>(<span>encoder_name: str, cfg: omegaconf.dictconfig.DictConfig)</span>
|
||
</code></dt>
|
||
<dd>
|
||
<details class="source">
|
||
<summary>
|
||
<span>Expand source code</span>
|
||
</summary>
|
||
<pre><code class="python">def get_encodec_autoencoder(encoder_name: str, cfg: omegaconf.DictConfig):
|
||
if encoder_name == "seanet":
|
||
kwargs = dict_from_config(getattr(cfg, "seanet"))
|
||
encoder_override_kwargs = kwargs.pop("encoder")
|
||
decoder_override_kwargs = kwargs.pop("decoder")
|
||
encoder_kwargs = {**kwargs, **encoder_override_kwargs}
|
||
decoder_kwargs = {**kwargs, **decoder_override_kwargs}
|
||
encoder = audiocraft.modules.SEANetEncoder(**encoder_kwargs)
|
||
decoder = audiocraft.modules.SEANetDecoder(**decoder_kwargs)
|
||
return encoder, decoder
|
||
else:
|
||
raise KeyError(f"Unexpected compression model {cfg.compression_model}")</code></pre>
|
||
</details>
|
||
<div class="desc"></div>
|
||
</dd>
|
||
<dt id="audiocraft.models.builders.get_jasco_model"><code class="name flex">
|
||
<span>def <span class="ident">get_jasco_model</span></span>(<span>cfg: omegaconf.dictconfig.DictConfig,<br>compression_model: <a title="audiocraft.models.encodec.CompressionModel" href="encodec.html#audiocraft.models.encodec.CompressionModel">CompressionModel</a> | None = None) ‑> <a title="audiocraft.models.flow_matching.FlowMatchingModel" href="flow_matching.html#audiocraft.models.flow_matching.FlowMatchingModel">FlowMatchingModel</a></span>
|
||
</code></dt>
|
||
<dd>
|
||
<details class="source">
|
||
<summary>
|
||
<span>Expand source code</span>
|
||
</summary>
|
||
<pre><code class="python">def get_jasco_model(cfg: omegaconf.DictConfig,
|
||
compression_model: tp.Optional[CompressionModel] = None) -> FlowMatchingModel:
|
||
kwargs = dict_from_config(getattr(cfg, "transformer_lm"))
|
||
attribute_dropout = dict_from_config(getattr(cfg, "attribute_dropout"))
|
||
cls_free_guidance = dict_from_config(getattr(cfg, "classifier_free_guidance"))
|
||
cfg_prob = cls_free_guidance["training_dropout"]
|
||
cfg_coef = cls_free_guidance["inference_coef"]
|
||
fuser = get_condition_fuser(cfg)
|
||
condition_provider = get_conditioner_provider(kwargs["dim"], cfg).to(cfg.device)
|
||
if JascoCondConst.DRM.value in condition_provider.conditioners: # use self_wav for drums
|
||
assert compression_model is not None
|
||
|
||
# use compression model for drums conditioning
|
||
condition_provider.conditioners.self_wav.compression_model = compression_model
|
||
condition_provider.conditioners.self_wav.compression_model.requires_grad_(False)
|
||
|
||
# downcast to jasco conditioning provider
|
||
seq_len = cfg.compression_model_framerate * cfg.dataset.segment_duration
|
||
chords_card = cfg.conditioners.chords.chords_emb.card if JascoCondConst.CRD.value in cfg.conditioners else -1
|
||
condition_provider = JascoConditioningProvider(device=condition_provider.device,
|
||
conditioners=condition_provider.conditioners,
|
||
chords_card=chords_card,
|
||
sequence_length=seq_len)
|
||
|
||
if len(fuser.fuse2cond["cross"]) > 0: # enforce cross-att programmatically
|
||
kwargs["cross_attention"] = True
|
||
|
||
kwargs.pop("n_q", None)
|
||
kwargs.pop("card", None)
|
||
|
||
return FlowMatchingModel(
|
||
condition_provider=condition_provider,
|
||
fuser=fuser,
|
||
cfg_dropout=cfg_prob,
|
||
cfg_coef=cfg_coef,
|
||
attribute_dropout=attribute_dropout,
|
||
dtype=getattr(torch, cfg.dtype),
|
||
device=cfg.device,
|
||
**kwargs,
|
||
).to(cfg.device)</code></pre>
|
||
</details>
|
||
<div class="desc"></div>
|
||
</dd>
|
||
<dt id="audiocraft.models.builders.get_lm_model"><code class="name flex">
|
||
<span>def <span class="ident">get_lm_model</span></span>(<span>cfg: omegaconf.dictconfig.DictConfig) ‑> <a title="audiocraft.models.lm.LMModel" href="lm.html#audiocraft.models.lm.LMModel">LMModel</a></span>
|
||
</code></dt>
|
||
<dd>
|
||
<details class="source">
|
||
<summary>
|
||
<span>Expand source code</span>
|
||
</summary>
|
||
<pre><code class="python">def get_lm_model(cfg: omegaconf.DictConfig) -> LMModel:
|
||
"""Instantiate a transformer LM."""
|
||
if cfg.lm_model in ["transformer_lm", "transformer_lm_magnet"]:
|
||
kwargs = dict_from_config(getattr(cfg, "transformer_lm"))
|
||
n_q = kwargs["n_q"]
|
||
q_modeling = kwargs.pop("q_modeling", None)
|
||
codebooks_pattern_cfg = getattr(cfg, "codebooks_pattern")
|
||
attribute_dropout = dict_from_config(getattr(cfg, "attribute_dropout"))
|
||
cls_free_guidance = dict_from_config(getattr(cfg, "classifier_free_guidance"))
|
||
cfg_prob, cfg_coef = (
|
||
cls_free_guidance["training_dropout"],
|
||
cls_free_guidance["inference_coef"],
|
||
)
|
||
fuser = get_condition_fuser(cfg)
|
||
condition_provider = get_conditioner_provider(kwargs["dim"], cfg).to(cfg.device)
|
||
if len(fuser.fuse2cond["cross"]) > 0: # enforce cross-att programmatically
|
||
kwargs["cross_attention"] = True
|
||
if codebooks_pattern_cfg.modeling is None:
|
||
assert (
|
||
q_modeling is not None
|
||
), "LM model should either have a codebook pattern defined or transformer_lm.q_modeling"
|
||
codebooks_pattern_cfg = omegaconf.OmegaConf.create(
|
||
{"modeling": q_modeling, "delay": {"delays": list(range(n_q))}}
|
||
)
|
||
|
||
pattern_provider = get_codebooks_pattern_provider(n_q, codebooks_pattern_cfg)
|
||
lm_class = MagnetLMModel if cfg.lm_model == "transformer_lm_magnet" else LMModel
|
||
return lm_class(
|
||
pattern_provider=pattern_provider,
|
||
condition_provider=condition_provider,
|
||
fuser=fuser,
|
||
cfg_dropout=cfg_prob,
|
||
cfg_coef=cfg_coef,
|
||
attribute_dropout=attribute_dropout,
|
||
dtype=getattr(torch, cfg.dtype),
|
||
device=cfg.device,
|
||
**kwargs,
|
||
).to(cfg.device)
|
||
else:
|
||
raise KeyError(f"Unexpected LM model {cfg.lm_model}")</code></pre>
|
||
</details>
|
||
<div class="desc"><p>Instantiate a transformer LM.</p></div>
|
||
</dd>
|
||
<dt id="audiocraft.models.builders.get_processor"><code class="name flex">
|
||
<span>def <span class="ident">get_processor</span></span>(<span>cfg, sample_rate: int = 24000)</span>
|
||
</code></dt>
|
||
<dd>
|
||
<details class="source">
|
||
<summary>
|
||
<span>Expand source code</span>
|
||
</summary>
|
||
<pre><code class="python">def get_processor(cfg, sample_rate: int = 24000):
|
||
sample_processor = SampleProcessor()
|
||
if cfg.use:
|
||
kw = dict(cfg)
|
||
kw.pop("use")
|
||
kw.pop("name")
|
||
if cfg.name == "multi_band_processor":
|
||
sample_processor = MultiBandProcessor(sample_rate=sample_rate, **kw)
|
||
return sample_processor</code></pre>
|
||
</details>
|
||
<div class="desc"></div>
|
||
</dd>
|
||
<dt id="audiocraft.models.builders.get_quantizer"><code class="name flex">
|
||
<span>def <span class="ident">get_quantizer</span></span>(<span>quantizer: str, cfg: omegaconf.dictconfig.DictConfig, dimension: int) ‑> <a title="audiocraft.quantization.base.BaseQuantizer" href="../quantization/base.html#audiocraft.quantization.base.BaseQuantizer">BaseQuantizer</a></span>
|
||
</code></dt>
|
||
<dd>
|
||
<details class="source">
|
||
<summary>
|
||
<span>Expand source code</span>
|
||
</summary>
|
||
<pre><code class="python">def get_quantizer(
|
||
quantizer: str, cfg: omegaconf.DictConfig, dimension: int
|
||
) -> qt.BaseQuantizer:
|
||
klass = {"no_quant": qt.DummyQuantizer, "rvq": qt.ResidualVectorQuantizer}[
|
||
quantizer
|
||
]
|
||
kwargs = dict_from_config(getattr(cfg, quantizer))
|
||
if quantizer != "no_quant":
|
||
kwargs["dimension"] = dimension
|
||
return klass(**kwargs)</code></pre>
|
||
</details>
|
||
<div class="desc"></div>
|
||
</dd>
|
||
<dt id="audiocraft.models.builders.get_watermark_model"><code class="name flex">
|
||
<span>def <span class="ident">get_watermark_model</span></span>(<span>cfg: omegaconf.dictconfig.DictConfig) ‑> <a title="audiocraft.models.watermark.WMModel" href="watermark.html#audiocraft.models.watermark.WMModel">WMModel</a></span>
|
||
</code></dt>
|
||
<dd>
|
||
<details class="source">
|
||
<summary>
|
||
<span>Expand source code</span>
|
||
</summary>
|
||
<pre><code class="python">def get_watermark_model(cfg: omegaconf.DictConfig) -> WMModel:
|
||
"""Build a WMModel based by audioseal. This requires audioseal to be installed"""
|
||
import audioseal
|
||
|
||
from .watermark import AudioSeal
|
||
|
||
# Builder encoder and decoder directly using audiocraft API to avoid cyclic import
|
||
assert hasattr(
|
||
cfg, "seanet"
|
||
), "Missing required `seanet` parameters in AudioSeal config"
|
||
encoder, decoder = get_encodec_autoencoder("seanet", cfg)
|
||
|
||
# Build message processor
|
||
kwargs = (
|
||
dict_from_config(getattr(cfg, "audioseal")) if hasattr(cfg, "audioseal") else {}
|
||
)
|
||
nbits = kwargs.get("nbits", 0)
|
||
hidden_size = getattr(cfg.seanet, "dimension", 128)
|
||
msg_processor = audioseal.MsgProcessor(nbits, hidden_size=hidden_size)
|
||
|
||
# Build detector using audioseal API
|
||
def _get_audioseal_detector():
|
||
# We don't need encoder and decoder params from seanet, remove them
|
||
seanet_cfg = dict_from_config(cfg.seanet)
|
||
seanet_cfg.pop("encoder")
|
||
seanet_cfg.pop("decoder")
|
||
detector_cfg = dict_from_config(cfg.detector)
|
||
|
||
typed_seanet_cfg = audioseal.builder.SEANetConfig(**seanet_cfg)
|
||
typed_detector_cfg = audioseal.builder.DetectorConfig(**detector_cfg)
|
||
_cfg = audioseal.builder.AudioSealDetectorConfig(
|
||
nbits=nbits, seanet=typed_seanet_cfg, detector=typed_detector_cfg
|
||
)
|
||
return audioseal.builder.create_detector(_cfg)
|
||
|
||
detector = _get_audioseal_detector()
|
||
generator = audioseal.AudioSealWM(
|
||
encoder=encoder, decoder=decoder, msg_processor=msg_processor
|
||
)
|
||
model = AudioSeal(generator=generator, detector=detector, nbits=nbits)
|
||
|
||
device = torch.device(getattr(cfg, "device", "cpu"))
|
||
dtype = getattr(torch, getattr(cfg, "dtype", "float32"))
|
||
return model.to(device=device, dtype=dtype)</code></pre>
|
||
</details>
|
||
<div class="desc"><p>Build a WMModel based by audioseal. This requires audioseal to be installed</p></div>
|
||
</dd>
|
||
<dt id="audiocraft.models.builders.get_wrapped_compression_model"><code class="name flex">
|
||
<span>def <span class="ident">get_wrapped_compression_model</span></span>(<span>compression_model: <a title="audiocraft.models.encodec.CompressionModel" href="encodec.html#audiocraft.models.encodec.CompressionModel">CompressionModel</a>,<br>cfg: omegaconf.dictconfig.DictConfig) ‑> <a title="audiocraft.models.encodec.CompressionModel" href="encodec.html#audiocraft.models.encodec.CompressionModel">CompressionModel</a></span>
|
||
</code></dt>
|
||
<dd>
|
||
<details class="source">
|
||
<summary>
|
||
<span>Expand source code</span>
|
||
</summary>
|
||
<pre><code class="python">def get_wrapped_compression_model(
|
||
compression_model: CompressionModel, cfg: omegaconf.DictConfig
|
||
) -> CompressionModel:
|
||
if hasattr(cfg, "interleave_stereo_codebooks"):
|
||
if cfg.interleave_stereo_codebooks.use:
|
||
kwargs = dict_from_config(cfg.interleave_stereo_codebooks)
|
||
kwargs.pop("use")
|
||
compression_model = InterleaveStereoCompressionModel(
|
||
compression_model, **kwargs
|
||
)
|
||
if hasattr(cfg, "compression_model_n_q"):
|
||
if cfg.compression_model_n_q is not None:
|
||
compression_model.set_num_codebooks(cfg.compression_model_n_q)
|
||
return compression_model</code></pre>
|
||
</details>
|
||
<div class="desc"></div>
|
||
</dd>
|
||
</dl>
|
||
</section>
|
||
<section>
|
||
</section>
|
||
</article>
|
||
<nav id="sidebar">
|
||
<div class="toc">
|
||
<ul></ul>
|
||
</div>
|
||
<ul id="index">
|
||
<li><h3>Super-module</h3>
|
||
<ul>
|
||
<li><code><a title="audiocraft.models" href="index.html">audiocraft.models</a></code></li>
|
||
</ul>
|
||
</li>
|
||
<li><h3><a href="#header-functions">Functions</a></h3>
|
||
<ul class="">
|
||
<li><code><a title="audiocraft.models.builders.get_codebooks_pattern_provider" href="#audiocraft.models.builders.get_codebooks_pattern_provider">get_codebooks_pattern_provider</a></code></li>
|
||
<li><code><a title="audiocraft.models.builders.get_compression_model" href="#audiocraft.models.builders.get_compression_model">get_compression_model</a></code></li>
|
||
<li><code><a title="audiocraft.models.builders.get_condition_fuser" href="#audiocraft.models.builders.get_condition_fuser">get_condition_fuser</a></code></li>
|
||
<li><code><a title="audiocraft.models.builders.get_conditioner_provider" href="#audiocraft.models.builders.get_conditioner_provider">get_conditioner_provider</a></code></li>
|
||
<li><code><a title="audiocraft.models.builders.get_debug_compression_model" href="#audiocraft.models.builders.get_debug_compression_model">get_debug_compression_model</a></code></li>
|
||
<li><code><a title="audiocraft.models.builders.get_debug_lm_model" href="#audiocraft.models.builders.get_debug_lm_model">get_debug_lm_model</a></code></li>
|
||
<li><code><a title="audiocraft.models.builders.get_diffusion_model" href="#audiocraft.models.builders.get_diffusion_model">get_diffusion_model</a></code></li>
|
||
<li><code><a title="audiocraft.models.builders.get_encodec_autoencoder" href="#audiocraft.models.builders.get_encodec_autoencoder">get_encodec_autoencoder</a></code></li>
|
||
<li><code><a title="audiocraft.models.builders.get_jasco_model" href="#audiocraft.models.builders.get_jasco_model">get_jasco_model</a></code></li>
|
||
<li><code><a title="audiocraft.models.builders.get_lm_model" href="#audiocraft.models.builders.get_lm_model">get_lm_model</a></code></li>
|
||
<li><code><a title="audiocraft.models.builders.get_processor" href="#audiocraft.models.builders.get_processor">get_processor</a></code></li>
|
||
<li><code><a title="audiocraft.models.builders.get_quantizer" href="#audiocraft.models.builders.get_quantizer">get_quantizer</a></code></li>
|
||
<li><code><a title="audiocraft.models.builders.get_watermark_model" href="#audiocraft.models.builders.get_watermark_model">get_watermark_model</a></code></li>
|
||
<li><code><a title="audiocraft.models.builders.get_wrapped_compression_model" href="#audiocraft.models.builders.get_wrapped_compression_model">get_wrapped_compression_model</a></code></li>
|
||
</ul>
|
||
</li>
|
||
</ul>
|
||
</nav>
|
||
</main>
|
||
<footer id="footer">
|
||
<p>Generated by <a href="https://pdoc3.github.io/pdoc" title="pdoc: Python API documentation generator"><cite>pdoc</cite> 0.11.5</a>.</p>
|
||
</footer>
|
||
</body>
|
||
</html>
|