datasets/v2.2.2/en/audio_process.html

<meta charset="utf-8" /><meta http-equiv="content-security-policy" content=""><meta name="hf:doc:metadata" content="{&quot;local&quot;:&quot;process-audio-data&quot;,&quot;sections&quot;:[{&quot;local&quot;:&quot;installation&quot;,&quot;title&quot;:&quot;Installation&quot;},{&quot;local&quot;:&quot;audio-datasets&quot;,&quot;title&quot;:&quot;Audio datasets&quot;},{&quot;local&quot;:&quot;resample&quot;,&quot;title&quot;:&quot;Resample&quot;},{&quot;local&quot;:&quot;map&quot;,&quot;title&quot;:&quot;Map&quot;}],&quot;title&quot;:&quot;Process audio data&quot;}" data-svelte="svelte-1phssyn">
	<link rel="modulepreload" href="/docs/datasets/v2.2.2/en/_app/assets/pages/__layout.svelte-efc77dbd.css">
	<link rel="modulepreload" href="/docs/datasets/v2.2.2/en/_app/start-0f8c1da7.js">
	<link rel="modulepreload" href="/docs/datasets/v2.2.2/en/_app/chunks/vendor-8138ceec.js">
	<link rel="modulepreload" href="/docs/datasets/v2.2.2/en/_app/chunks/paths-4b3c6e7e.js">
	<link rel="modulepreload" href="/docs/datasets/v2.2.2/en/_app/pages/__layout.svelte-efb8e839.js">
	<link rel="modulepreload" href="/docs/datasets/v2.2.2/en/_app/pages/audio_process.mdx-4288cdc1.js">
	<link rel="modulepreload" href="/docs/datasets/v2.2.2/en/_app/chunks/Tip-12722dfc.js">
	<link rel="modulepreload" href="/docs/datasets/v2.2.2/en/_app/chunks/IconCopyLink-2dd3a6ac.js">
	<link rel="modulepreload" href="/docs/datasets/v2.2.2/en/_app/chunks/CodeBlock-fc89709f.js"> 





<h1 class="relative group"><a id="process-audio-data" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#process-audio-data"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a>
	<span>Process audio data
	</span></h1>

<p>🤗 Datasets supports an <a href="/docs/datasets/v2.2.2/en/package_reference/main_classes#datasets.Audio">Audio</a> feature, enabling users to load and process raw audio files for training. This guide will show you how to:</p>
<ul><li>Load your own custom audio dataset.</li>
<li>Resample audio files.</li>
<li>Use <a href="/docs/datasets/v2.2.2/en/package_reference/main_classes#datasets.Dataset.map">Dataset.map()</a> with audio files.</li></ul>
<h2 class="relative group"><a id="installation" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#installation"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a>
	<span>Installation
	</span></h2>

<p>The <a href="/docs/datasets/v2.2.2/en/package_reference/main_classes#datasets.Audio">Audio</a> feature should be installed as an extra dependency in 🤗 Datasets. Install the <a href="/docs/datasets/v2.2.2/en/package_reference/main_classes#datasets.Audio">Audio</a> feature (and its dependencies) with pip:</p>

	<div class="code-block relative"><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5   text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg>
	<div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div>
	Copied</div></button></div>
	<pre><!-- HTML_TAG_START -->pip install datasets[audio]<!-- HTML_TAG_END --></pre></div>


<div class="course-tip course-tip-orange bg-gradient-to-br dark:bg-gradient-to-r before:border-orange-500 dark:before:border-orange-800 from-orange-50 dark:from-gray-900 to-white dark:to-gray-950 border border-orange-50 text-orange-700 dark:text-gray-400"><p>On Linux, non-Python dependency on <code>libsndfile</code> package must be installed manually, using your distribution package manager, for example:</p>

	<div class="code-block relative"><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5   text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg>
	<div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div>
	Copied</div></button></div>
	<pre><!-- HTML_TAG_START -->sudo apt-get install libsndfile1<!-- HTML_TAG_END --></pre></div></div>
<p>To support loading audio datasets containing MP3 files, users should additionally install <a href="https://pytorch.org/audio/stable/index.html" rel="nofollow">torchaudio</a>, so that audio data is handled with high performance.</p>

	<div class="code-block relative"><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5   text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg>
	<div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div>
	Copied</div></button></div>
	<pre><!-- HTML_TAG_START -->pip install torchaudio<!-- HTML_TAG_END --></pre></div>


<div class="course-tip course-tip-orange bg-gradient-to-br dark:bg-gradient-to-r before:border-orange-500 dark:before:border-orange-800 from-orange-50 dark:from-gray-900 to-white dark:to-gray-950 border border-orange-50 text-orange-700 dark:text-gray-400"><p>torchaudio’s <code>sox_io</code> <a href="https://pytorch.org/audio/stable/backend.html#" rel="nofollow">backend</a> supports decoding <code>mp3</code> files. Unfortunately, the <code>sox_io</code> backend is only available on Linux/macOS, and is not supported by Windows.</p></div>
<p>Then you can load an audio dataset the same way you would load a text dataset. For example, load the <a href="https://huggingface.co/datasets/common_voice" rel="nofollow">Common Voice</a> dataset with the Turkish configuration:</p>

	<div class="code-block relative"><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5   text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg>
	<div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div>
	Copied</div></button></div>
	<pre><!-- HTML_TAG_START --><span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-keyword">from</span> datasets <span class="hljs-keyword">import</span> load_dataset, load_metric, Audio
<span class="hljs-meta">&gt;&gt;&gt; </span>common_voice = load_dataset(<span class="hljs-string">&quot;common_voice&quot;</span>, <span class="hljs-string">&quot;tr&quot;</span>, split=<span class="hljs-string">&quot;train&quot;</span>)<!-- HTML_TAG_END --></pre></div>
<h2 class="relative group"><a id="audio-datasets" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#audio-datasets"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a>
	<span>Audio datasets
	</span></h2>

<p>Audio datasets commonly have an <code>audio</code> and <code>path</code> or <code>file</code> column.</p>
<p><code>audio</code> is the actual audio file that is loaded and resampled on-the-fly upon calling it.</p>

	<div class="code-block relative"><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5   text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg>
	<div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div>
	Copied</div></button></div>
	<pre><!-- HTML_TAG_START --><span class="hljs-meta">&gt;&gt;&gt; </span>common_voice[<span class="hljs-number">0</span>][<span class="hljs-string">&quot;audio&quot;</span>]
{<span class="hljs-string">&#x27;array&#x27;</span>: array([ <span class="hljs-number">0.0000000e+00</span>,  <span class="hljs-number">0.0000000e+00</span>,  <span class="hljs-number">0.0000000e+00</span>, ...,
    -<span class="hljs-number">8.8930130e-05</span>, -<span class="hljs-number">3.8027763e-05</span>, -<span class="hljs-number">2.9146671e-05</span>], dtype=float32),
<span class="hljs-string">&#x27;path&#x27;</span>: <span class="hljs-string">&#x27;/root/.cache/huggingface/datasets/downloads/extracted/05be0c29807a73c9b099873d2f5975dae6d05e9f7d577458a2466ecb9a2b0c6b/cv-corpus-6.1-2020-12-11/tr/clips/common_voice_tr_21921195.mp3&#x27;</span>,
<span class="hljs-string">&#x27;sampling_rate&#x27;</span>: <span class="hljs-number">48000</span>}<!-- HTML_TAG_END --></pre></div>
<p>When you access an audio file, it is automatically decoded and resampled. Generally, you should query an audio file like: <code>common_voice[0][&quot;audio&quot;]</code>. If you query an audio file with <code>common_voice[&quot;audio&quot;][0]</code> instead, <strong>all</strong> the audio files in your dataset will be decoded and resampled. This process can take a long time if you have a large dataset.</p>
<p><code>path</code> or <code>file</code> is an absolute path to an audio file.</p>

	<div class="code-block relative"><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5   text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg>
	<div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div>
	Copied</div></button></div>
	<pre><!-- HTML_TAG_START --><span class="hljs-meta">&gt;&gt;&gt; </span>common_voice[<span class="hljs-number">0</span>][<span class="hljs-string">&quot;path&quot;</span>]
/root/.cache/huggingface/datasets/downloads/extracted/05be0c29807a73c9b099873d2f5975dae6d05e9f7d577458a2466ecb9a2b0c6b/cv-corpus-<span class="hljs-number">6.1</span>-<span class="hljs-number">2020</span>-<span class="hljs-number">12</span>-<span class="hljs-number">11</span>/tr/clips/common_voice_tr_21921195.mp3<!-- HTML_TAG_END --></pre></div>
<p>The <code>path</code> is useful if you want to load your own audio dataset. In this case, provide a column of audio file paths to <a href="/docs/datasets/v2.2.2/en/package_reference/main_classes#datasets.Dataset.cast_column">Dataset.cast_column()</a>:</p>

	<div class="code-block relative"><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5   text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg>
	<div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div>
	Copied</div></button></div>
	<pre><!-- HTML_TAG_START --><span class="hljs-meta">&gt;&gt;&gt; </span>my_audio_dataset = my_audio_dataset.cast_column(<span class="hljs-string">&quot;paths_to_my_audio_files&quot;</span>, Audio())<!-- HTML_TAG_END --></pre></div>
<h2 class="relative group"><a id="resample" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#resample"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a>
	<span>Resample
	</span></h2>

<p>Some models expect the audio data to have a certain sampling rate due to how the model was pretrained. For example, the <a href="https://huggingface.co/facebook/wav2vec2-large-xlsr-53" rel="nofollow">XLSR-Wav2Vec2</a> model expects the input to have a sampling rate of 16kHz, but an audio file from the Common Voice dataset has a sampling rate of 48kHz. You can use <a href="/docs/datasets/v2.2.2/en/package_reference/main_classes#datasets.Dataset.cast_column">Dataset.cast_column()</a> to downsample the sampling rate to 16kHz:</p>

	<div class="code-block relative"><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5   text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg>
	<div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div>
	Copied</div></button></div>
	<pre><!-- HTML_TAG_START --><span class="hljs-meta">&gt;&gt;&gt; </span>common_voice = common_voice.cast_column(<span class="hljs-string">&quot;audio&quot;</span>, Audio(sampling_rate=<span class="hljs-number">16_000</span>))<!-- HTML_TAG_END --></pre></div>
<p>The next time you load the audio file, the <a href="/docs/datasets/v2.2.2/en/package_reference/main_classes#datasets.Audio">Audio</a> feature will load and resample it to 16kHz:</p>

	<div class="code-block relative"><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5   text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg>
	<div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div>
	Copied</div></button></div>
	<pre><!-- HTML_TAG_START --><span class="hljs-meta">&gt;&gt;&gt; </span>common_voice_train[<span class="hljs-number">0</span>][<span class="hljs-string">&quot;audio&quot;</span>]
{<span class="hljs-string">&#x27;array&#x27;</span>: array([ <span class="hljs-number">0.0000000e+00</span>,  <span class="hljs-number">0.0000000e+00</span>,  <span class="hljs-number">0.0000000e+00</span>, ...,
-<span class="hljs-number">7.4556941e-05</span>, -<span class="hljs-number">1.4621433e-05</span>, -<span class="hljs-number">5.7861507e-05</span>], dtype=float32),
<span class="hljs-string">&#x27;path&#x27;</span>: <span class="hljs-string">&#x27;/root/.cache/huggingface/datasets/downloads/extracted/05be0c29807a73c9b099873d2f5975dae6d05e9f7d577458a2466ecb9a2b0c6b/cv-corpus-6.1-2020-12-11/tr/clips/common_voice_tr_21921195.mp3&#x27;</span>,
<span class="hljs-string">&#x27;sampling_rate&#x27;</span>: <span class="hljs-number">16000</span>}<!-- HTML_TAG_END --></pre></div>
<div class="flex justify-center"><img class="block dark:hidden" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/datasets/resample.gif">
    <img class="hidden dark:block" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/datasets/resample-dark.gif"></div>
<h2 class="relative group"><a id="map" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#map"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a>
	<span>Map
	</span></h2>

<p>Just like text datasets, you can apply a preprocessing function over an entire dataset with <a href="/docs/datasets/v2.2.2/en/package_reference/main_classes#datasets.Dataset.map">Dataset.map()</a>, which is useful for preprocessing all of your audio data at once. Start with a <a href="https://huggingface.co/models?pipeline_tag=automatic-speech-recognition&sort=downloads" rel="nofollow">speech recognition model</a> of your choice, and load a <code>processor</code> object that contains:</p>
<ol><li><p>A feature extractor to convert the speech signal to the model’s input format. Every speech recognition model on the 🤗 <a href="https://huggingface.co/models?pipeline_tag=automatic-speech-recognition&sort=downloads" rel="nofollow">Hub</a> contains a predefined feature extractor that can be easily loaded with <code>AutoFeatureExtractor.from_pretrained(...)</code>.</p></li>
<li><p>A tokenizer to convert the model’s output format to text. Fine-tuned speech recognition models, such as <a href="https://huggingface.co/facebook/wav2vec2-base-960h" rel="nofollow">facebook/wav2vec2-base-960h</a>, contain a predefined tokenizer that can be easily loaded with <code>AutoTokenizer.from_pretrained(...)</code>.</p>
<p>For pretrained speech recognition models, such as <a href="https://huggingface.co/facebook/wav2vec2-large-xlsr-53" rel="nofollow">facebook/wav2vec2-large-xlsr-53</a>, a tokenizer needs to be created from the target text as explained <a href="https://huggingface.co/blog/fine-tune-wav2vec2-english" rel="nofollow">here</a>. The following example demonstrates how to load a feature extractor, tokenizer and processor for a pretrained speech recognition model:</p></li></ol>

	<div class="code-block relative"><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5   text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg>
	<div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div>
	Copied</div></button></div>
	<pre><!-- HTML_TAG_START --><span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> AutoTokenizer, AutoFeatureExtractor, Wav2Vec2Processor
<span class="hljs-meta">&gt;&gt;&gt; </span>model_checkpoint = <span class="hljs-string">&quot;facebook/wav2vec2-large-xlsr-53&quot;</span>
<span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-comment"># after defining a vocab.json file you can instantiate a tokenizer object:</span>
<span class="hljs-meta">&gt;&gt;&gt; </span>tokenizer = AutoTokenizer(<span class="hljs-string">&quot;./vocab.json&quot;</span>, unk_token=<span class="hljs-string">&quot;[UNK]&quot;</span>, pad_token=<span class="hljs-string">&quot;[PAD]&quot;</span>, word_delimiter_token=<span class="hljs-string">&quot;|&quot;</span>)
<span class="hljs-meta">&gt;&gt;&gt; </span>feature_extractor = AutoFeatureExtractor.from_pretrained(model_checkpoint)
<span class="hljs-meta">&gt;&gt;&gt; </span>processor = Wav2Vec2Processor.from_pretrained(feature_extractor=feature_extractor, tokenizer=tokenizer)<!-- HTML_TAG_END --></pre></div>
<p>For fine-tuned speech recognition models, you can simply load a predefined <code>processor</code> object with:</p>

	<div class="code-block relative"><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5   text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg>
	<div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div>
	Copied</div></button></div>
	<pre><!-- HTML_TAG_START --><span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> Wav2Vec2Processor
<span class="hljs-meta">&gt;&gt;&gt; </span>processor = Wav2Vec2Processor.from_pretrained(<span class="hljs-string">&quot;facebook/wav2vec2-base-960h&quot;</span>)<!-- HTML_TAG_END --></pre></div>
<p>Make sure to include the <code>audio</code> key in your preprocessing function when you call <a href="/docs/datasets/v2.2.2/en/package_reference/main_classes#datasets.Dataset.map">Dataset.map()</a> so that you are actually resampling the audio data:</p>

	<div class="code-block relative"><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5   text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg>
	<div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div>
	Copied</div></button></div>
	<pre><!-- HTML_TAG_START --><span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-keyword">def</span> <span class="hljs-title function_">prepare_dataset</span>(<span class="hljs-params">batch</span>):
<span class="hljs-meta">... </span>    audio = batch[<span class="hljs-string">&quot;audio&quot;</span>]
<span class="hljs-meta">... </span>    batch[<span class="hljs-string">&quot;input_values&quot;</span>] = processor(audio[<span class="hljs-string">&quot;array&quot;</span>], sampling_rate=audio[<span class="hljs-string">&quot;sampling_rate&quot;</span>]).input_values[<span class="hljs-number">0</span>]
<span class="hljs-meta">... </span>    batch[<span class="hljs-string">&quot;input_length&quot;</span>] = <span class="hljs-built_in">len</span>(batch[<span class="hljs-string">&quot;input_values&quot;</span>])
<span class="hljs-meta">... </span>    <span class="hljs-keyword">with</span> processor.as_target_processor():
<span class="hljs-meta">... </span>        batch[<span class="hljs-string">&quot;labels&quot;</span>] = processor(batch[<span class="hljs-string">&quot;sentence&quot;</span>]).input_ids
<span class="hljs-meta">... </span>    <span class="hljs-keyword">return</span> batch
<span class="hljs-meta">&gt;&gt;&gt; </span>common_voice_train = common_voice_train.<span class="hljs-built_in">map</span>(prepare_dataset, remove_columns=common_voice_train.column_names)<!-- HTML_TAG_END --></pre></div>


		<script type="module" data-hydrate="1vul9gx">
		import { start } from "/docs/datasets/v2.2.2/en/_app/start-0f8c1da7.js";
		start({
			target: document.querySelector('[data-hydrate="1vul9gx"]').parentNode,
			paths: {"base":"/docs/datasets/v2.2.2/en","assets":"/docs/datasets/v2.2.2/en"},
			session: {},
			route: false,
			spa: false,
			trailing_slash: "never",
			hydrate: {
				status: 200,
				error: null,
				nodes: [
					import("/docs/datasets/v2.2.2/en/_app/pages/__layout.svelte-efb8e839.js"),
						import("/docs/datasets/v2.2.2/en/_app/pages/audio_process.mdx-4288cdc1.js")
				],
				params: {}
			}
		});
	</script>