Add torchaudio, torchcodec and datasets ways to load the audio

README.md

@@ -102,10 +102,9 @@ model = AutoModelForAudioClassification.from_pretrained(
     repo, subfolder="int8").eval()          # or "fp8", "int4"
 fe = AutoFeatureExtractor.from_pretrained(repo, subfolder="int8")
 
-# librosa resamples and downmixes. Do not skip it: RESD itself ships at
-# 44.1 kHz, and handing the model 44.1 kHz audio while telling the
-# extractor it is 16 kHz stretches time 2.8x and silently changes the
-# answer rather than raising.
+# Resample to 16 kHz. Do not skip it: RESD itself ships at 44.1 kHz,
+# and handing the model 44.1 kHz audio while telling the extractor it
+# is 16 kHz stretches time 2.8x and silently changes the answer.
 wav, _ = librosa.load("clip.wav", sr=16000, mono=True)
 x = fe(wav, sampling_rate=16000, return_tensors="pt", padding=True)
 with torch.no_grad():
@@ -113,6 +112,27 @@ with torch.no_grad():
 print({model.config.id2label[i]: round(p.item(), 3) for i, p in enumerate(probs)})
 ```
 
+<details><summary>Loading the audio without librosa</summary>
+
+```python
+# torchaudio
+import torchaudio
+wav, sr = torchaudio.load("clip.wav")
+wav = torchaudio.functional.resample(wav, sr, 16000).mean(0).numpy()
+
+# torchcodec, the newer decoder
+from torchcodec.decoders import AudioDecoder
+wav = AudioDecoder("clip.wav", sample_rate=16000).get_all_samples().data.mean(0).numpy()
+
+# straight from the dataset, which resamples on the column
+from datasets import load_dataset, Audio
+ds = load_dataset("Aniemore/resd", split="test")
+ds = ds.cast_column("speech", Audio(sampling_rate=16000))
+wav = ds[0]["speech"]["array"]
+```
+
+</details>
+
 ## What is quantized
 
 Only `nn.Linear` inside the encoder blocks &mdash; that is where 98.6% of the weight mass sits. Left in fp32: the convolutional feature extractor, the positional convolution, the layer norms, the projector and the classifier. The classifier is seven rows wide, so quantizing it would save nothing and put rounding error straight onto the logits.