first app

app.py

@@ -0,0 +1,27 @@
+import tensorflow as tf
+from transformers import Speech2TextProcessor, TFSpeech2TextForConditionalGeneration
+from datasets import load_dataset
+import soundfile as sf
+
+model = TFSpeech2TextForConditionalGeneration.from_pretrained(
+    "facebook/s2t-small-librispeech-asr", from_pt=True
+)
+processor = Speech2TextProcessor.from_pretrained("facebook/s2t-small-librispeech-asr")
+
+
+def map_to_array(batch):
+    speech, _ = sf.read(batch["file"])
+    batch["speech"] = speech
+    return batch
+
+
+ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
+ds = ds.map(map_to_array)
+ds.set_format(type="tf")
+
+input_features = processor(
+    ds["speech"][0], sampling_rate=16000, return_tensors="tf"
+).input_features  # Batch size 1
+generated_ids = model.generate(input_features)
+
+transcription = processor.batch_decode(generated_ids)