Code block showing how with JavaScript and HuggingFace API, you can easily convert audio feedback from the customer to text and then understand its sentiment.

audio-to-text-to-sentiment.js

const {
  HUGGINGFACE_API_KEY,
} = process.env;

async function fetchHuggingFaceModel(model, data) {
  const response = await fetch(
    `https://api-inference.huggingface.co/models/${model}`,
    {
      headers: { Authorization: `Bearer ${HUGGINGFACE_API_KEY}` },
      method: 'POST',
      body: data,
      duplex: 'half',
    }
  );
  return response.json();
}

async function getTextFromSource(model, fileSource) {
  const getResponse = await fetch(fileSource);
  return fetchHuggingFaceModel(model, getResponse.body);
}

(async () => {
  // extracting audio to text
  const audioToText = await getTextFromSource(
    'openai/whisper-medium',
    'https://drive.google.com/uc?export=download&id=12l9vffZiE5ui4dLWADf9CmctwDH8KAN9'
  );
  // audioToText.text = `I really enjoyed visiting this tech conference, we are developers.
  //   There have been lots of great people to collaborate with and to speak with them.`

  // extract sentiment from text
  const sentiments = await fetchHuggingFaceModel(
    'lxyuan/distilbert-base-multilingual-cased-sentiments-student',
    JSON.stringify({
      inputs: audioToText.text,
    })
  );

  console.log('results', audioToText, sentiments);
  // results {
  //   text: ' I really enjoyed visiting this tech conference, we are developers. There have been lots of great people to collaborate with and to speak with them.'
  // } [
  //   [
  //     { label: 'positive', score: 0.8829012513160706 },
  //     { label: 'neutral', score: 0.06260993331670761 },
  //     { label: 'negative', score: 0.05448877066373825 }
  //   ]
  // ]
})();

Shared on twitter: https://x.com/AndrewRedUK/status/1753752365847441752?s=20