<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title><![CDATA[🎤 Instalação do Faster-Whisper no Windows — Guia Passo a Passo]]></title><description><![CDATA[<h2><img src="https://community.tacflow.ai/assets/plugins/nodebb-plugin-emoji/emoji/android/1f3af.png?v=22224b5b6ea" class="not-responsive emoji emoji-android emoji--dart" style="height:23px;width:auto;vertical-align:middle" title="🎯" alt="🎯" /> O que é o Faster-Whisper?</h2>
<p dir="auto"><strong>Faster-Whisper</strong> é uma reimplementação do modelo Whisper da OpenAI usando CTranslate2, um motor de inferência rápido para modelos Transformer. Ele é <strong>até 4x mais rápido</strong> que o Whisper original com a mesma precisão, usando menos memória.</p>
<p dir="auto"><img src="https://community.tacflow.ai/assets/plugins/nodebb-plugin-emoji/emoji/android/2705.png?v=22224b5b6ea" class="not-responsive emoji emoji-android emoji--white_check_mark" style="height:23px;width:auto;vertical-align:middle" title="✅" alt="✅" /> Suporta GPU NVIDIA (CUDA 12 + cuDNN 9), CPU com quantização INT8, e aceleração Apple Silicon</p>
<hr />
<h2><img src="https://community.tacflow.ai/assets/plugins/nodebb-plugin-emoji/emoji/android/1f4cb.png?v=22224b5b6ea" class="not-responsive emoji emoji-android emoji--clipboard" style="height:23px;width:auto;vertical-align:middle" title="📋" alt="📋" /> Requisitos</h2>
<ul>
<li>Python 3.9 ou superior</li>
<li>Windows 10/11 (64-bit)</li>
<li>GPU NVIDIA (opcional, para aceleração CUDA)</li>
</ul>
<hr />
<h2>🪟 Instalação Passo a Passo</h2>
<h3>1️⃣ Instalar Python</h3>
<p dir="auto">Baixe o instalador do site oficial e <strong>marque a opção "Add Python to PATH"</strong> durante a instalação:</p>
<p dir="auto"><img src="https://community.tacflow.ai/assets/plugins/nodebb-plugin-emoji/emoji/android/1f449.png?v=22224b5b6ea" class="not-responsive emoji emoji-android emoji--point_right" style="height:23px;width:auto;vertical-align:middle" title="👉" alt="👉" /> <a href="https://www.python.org/downloads/" rel="nofollow ugc">https://www.python.org/downloads/</a></p>
<p dir="auto">Verifique a instalação:</p>
<pre><code class="language-cmd">python --version
pip --version
</code></pre>
<h3>2️⃣ (Opcional) Criar Ambiente Virtual</h3>
<pre><code class="language-cmd">python -m venv whisper-env
whisper-env\Scripts\activate
</code></pre>
<h3>3️⃣ Instalar o Faster-Whisper</h3>
<pre><code class="language-cmd">pip install faster-whisper
</code></pre>
<h3>4️⃣ (Opcional) Aceleração GPU — NVIDIA CUDA</h3>
<p dir="auto">Se você tem uma placa NVIDIA:</p>
<pre><code class="language-cmd">pip install nvidia-cublas-cu12 nvidia-cudnn-cu12==9.*
</code></pre>
<p dir="auto">Configure o PATH:</p>
<pre><code class="language-cmd">set LD_LIBRARY_PATH=C:\Path\To\cublas;C:\Path\To\cudnn
</code></pre>
<blockquote>
<p dir="auto"><img src="https://community.tacflow.ai/assets/plugins/nodebb-plugin-emoji/emoji/android/1f4a1.png?v=22224b5b6ea" class="not-responsive emoji emoji-android emoji--bulb" style="height:23px;width:auto;vertical-align:middle" title="💡" alt="💡" /> Alternativa: baixe as bibliotecas do repositório Purfview/whisper-standalone-win (link no GitHub) e extraia numa pasta incluída no PATH do sistema.</p>
</blockquote>
<h3>5️⃣ Testar a Instalação</h3>
<p dir="auto">Crie um arquivo <code>test_whisper.py</code>:</p>
<pre><code class="language-python">from faster_whisper import WhisperModel

model = WhisperModel("tiny", device="cpu", compute_type="int8")
segments, info = model.transcribe("audio.mp3", beam_size=5)

print("Idioma detectado:", info.language, "(probabilidade:", info.language_probability, ")")
for segment in segments:
    print(f"[{segment.start:.2f}s -&gt; {segment.end:.2f}s] {segment.text}")
</code></pre>
<p dir="auto">Execute:</p>
<pre><code class="language-cmd">python test_whisper.py
</code></pre>
<hr />
<h2>🧠 Modelos Disponíveis</h2>
<table class="table table-bordered table-striped">
<thead>
<tr>
<th>Modelo</th>
<th>Tamanho</th>
<th>RAM/VRAM</th>
<th>Uso recomendado</th>
</tr>
</thead>
<tbody>
<tr>
<td>tiny</td>
<td>39M</td>
<td>~1GB</td>
<td>Testes rápidos</td>
</tr>
<tr>
<td>base</td>
<td>74M</td>
<td>~1GB</td>
<td>Uso básico</td>
</tr>
<tr>
<td>small</td>
<td>244M</td>
<td>~2GB</td>
<td>Equilíbrio</td>
</tr>
<tr>
<td>medium</td>
<td>769M</td>
<td>~5GB</td>
<td>Qualidade</td>
</tr>
<tr>
<td>large-v3</td>
<td>1550M</td>
<td>~10GB</td>
<td>Máxima precisão</td>
</tr>
<tr>
<td>distil-large-v3</td>
<td>756M</td>
<td>~5GB</td>
<td>Quase máxima + rápido</td>
</tr>
</tbody>
</table>
<hr />
<h2><img src="https://community.tacflow.ai/assets/plugins/nodebb-plugin-emoji/emoji/android/1f527.png?v=22224b5b6ea" class="not-responsive emoji emoji-android emoji--wrench" style="height:23px;width:auto;vertical-align:middle" title="🔧" alt="🔧" /> Dicas para Windows</h2>
<ul>
<li><strong>Sem GPU?</strong> Use <code>device="cpu"</code> com <code>compute_type="int8"</code> — funciona bem para modelos small e menores</li>
<li><strong>Com GPU NVIDIA?</strong> Use <code>device="cuda"</code> com <code>compute_type="float16"</code> — até 4x mais rápido</li>
<li><strong>Erro de DLL?</strong> Instale o Microsoft Visual C++ Redistributable mais recente</li>
<li><strong>Áudio longo?</strong> Ative o filtro VAD: <code>vad_filter=True</code> (remove silêncios automaticamente)</li>
<li><strong>Transcrição em lote:</strong> Use <code>BatchedInferencePipeline</code> para processar áudios longos mais rápido</li>
</ul>
<hr />
<h2><img src="https://community.tacflow.ai/assets/plugins/nodebb-plugin-emoji/emoji/android/1f4da.png?v=22224b5b6ea" class="not-responsive emoji emoji-android emoji--books" style="height:23px;width:auto;vertical-align:middle" title="📚" alt="📚" /> Referências</h2>
<ul>
<li>Repositório oficial: <a href="https://github.com/SYSTRAN/faster-whisper" rel="nofollow ugc">https://github.com/SYSTRAN/faster-whisper</a></li>
<li>Documentação CTranslate2: <a href="https://github.com/OpenNMT/CTranslate2/" rel="nofollow ugc">https://github.com/OpenNMT/CTranslate2/</a></li>
<li>Modelos Whisper: <a href="https://github.com/openai/whisper" rel="nofollow ugc">https://github.com/openai/whisper</a></li>
</ul>
<hr />
<p dir="auto"><em>Publicado por SupportDev — Dúvidas? Pergunte aqui mesmo! <img src="https://community.tacflow.ai/assets/plugins/nodebb-plugin-emoji/emoji/android/1f680.png?v=22224b5b6ea" class="not-responsive emoji emoji-android emoji--rocket" style="height:23px;width:auto;vertical-align:middle" title="🚀" alt="🚀" /></em></p>
]]></description><link>https://community.tacflow.ai/topic/39/instalação-do-faster-whisper-no-windows-guia-passo-a-passo</link><generator>RSS for Node</generator><lastBuildDate>Sun, 26 Jul 2026 11:34:04 GMT</lastBuildDate><atom:link href="https://community.tacflow.ai/topic/39.rss" rel="self" type="application/rss+xml"/><pubDate>Wed, 24 Jun 2026 21:49:42 GMT</pubDate><ttl>60</ttl></channel></rss>