pixeltable 0.2.23__py3-none-any.whl → 0.2.24__py3-none-any.whl

pixeltable/__version__.py

@@ -1,3 +1,3 @@
 # These version placeholders will be replaced during build.
-__version__ = "0.2.23"
-__version_tuple__ = (0, 2, 23)
+__version__ = "0.2.24"
+__version_tuple__ = (0, 2, 24)

pixeltable/functions/huggingface.py

@@ -426,12 +426,10 @@ def speech2text_for_conditional_generation(
     env.Env.get().require_package('torchaudio')
     env.Env.get().require_package('sentencepiece')
     device = resolve_torch_device('auto', allow_mps=False)  # Doesn't seem to work on 'mps'; use 'cpu' instead
-    import librosa
     import torch
+    import torchaudio  # type: ignore[import-untyped]
     from transformers import Speech2TextForConditionalGeneration, Speech2TextProcessor
 
-    # facebook/s2t-small-librispeech-asr
-    # facebook/s2t-small-mustc-en-fr-st
     model = _lookup_model(model_id, Speech2TextForConditionalGeneration.from_pretrained, device=device)
     processor = _lookup_processor(model_id, Speech2TextProcessor.from_pretrained)
     assert isinstance(processor, Speech2TextProcessor)
@@ -445,12 +443,23 @@ def speech2text_for_conditional_generation(
 
     # Get the model's sampling rate. Default to 16 kHz (the standard) if not in config
     model_sampling_rate = getattr(model.config, 'sampling_rate', 16_000)
-    waveform, sampling_rate = librosa.load(audio, sr=model_sampling_rate, mono=True)
+
+    waveform, sampling_rate = torchaudio.load(audio)
+
+    # Resample to the model's sampling rate, if necessary
+    if sampling_rate != model_sampling_rate:
+        waveform = torchaudio.transforms.Resample(sampling_rate, model_sampling_rate)(waveform)
+
+    # Average the channels to get a single-channel waveform as a 1D tensor (if the original waveform is already
+    # mono, this will simply squeeze the tensor)
+    assert waveform.dim() == 2
+    waveform = torch.mean(waveform, dim=0)
+    assert waveform.dim() == 1
 
     with torch.no_grad():
         inputs = processor(
             waveform,
-            sampling_rate=sampling_rate,
+            sampling_rate=model_sampling_rate,
             return_tensors='pt'
         )
         generated_ids = model.generate(**inputs.to(device), forced_bos_token_id=forced_bos_token_id).to('cpu')

{pixeltable-0.2.23.dist-info → pixeltable-0.2.24.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: pixeltable
-Version: 0.2.23
+Version: 0.2.24
 Summary: Pixeltable: The Multimodal AI Data Plane
 Author: Pixeltable, Inc.
 Author-email: contact@pixeltable.com

{pixeltable-0.2.23.dist-info → pixeltable-0.2.24.dist-info}/RECORD

@@ -1,5 +1,5 @@
 pixeltable/__init__.py,sha256=gv2jvZ7H5tEjLear10E7hSH9jF5Mw6iSeryvRp88bOE,1391
-pixeltable/__version__.py,sha256=7KY-iWvvGLR_f-tQMjxiOTfMCj-z3KM_23EsiPZSTl8,114
+pixeltable/__version__.py,sha256=w8n5Ad700zSFGVbR5bQpPtdgGwCXcl22svUmysoPXE0,114
 pixeltable/catalog/__init__.py,sha256=Ar6_F_6C7tkznIlCPBHVHDop5YssBDjKQr2NPQ21QCI,484
 pixeltable/catalog/catalog.py,sha256=tyDyI5wQw7vV6_FChrp9qgGCRClcjiSdW3eygYT0p9s,7849
 pixeltable/catalog/column.py,sha256=ezeKoGl6aBTzSZBihDA6vdETcvyCguAD4OsmrxWs73A,9595
@@ -74,7 +74,7 @@ pixeltable/functions/anthropic.py,sha256=P1E5o4-8QP1LTIUsWVgo_wMJ4WOnxtXUUXuFWUa
 pixeltable/functions/audio.py,sha256=7213nTnqKJ6vM9kalaoJ283OwX5SGEJN10vDhaRNZ6E,644
 pixeltable/functions/fireworks.py,sha256=qwFC_eIaDs-glxyJ_IVXaNGkpgPzeRsQ_SdpzueBxq0,2605
 pixeltable/functions/globals.py,sha256=pCFX2a_N87SwG9GxyPjSOC3TVMowMB6XIHSWKfFOuGE,3917
-pixeltable/functions/huggingface.py,sha256=hcScloxprIuyf89_bSnQi-N3VZRQosmZfBhmDJF_eTc,20814
+pixeltable/functions/huggingface.py,sha256=s5KmOfi9-TOYyrL1Wv-voKP7ykkUN7LlLAA_uo01UQc,21210
 pixeltable/functions/image.py,sha256=3Qm4ybAT_o4YUl3bzhEXy8dKOwgZ7RCUV-ky-dbL_jc,13836
 pixeltable/functions/json.py,sha256=ehCnBA5WiIl-crV9PFVgmxrsWsiO8FpRs9LDwcSpLa4,879
 pixeltable/functions/llama_cpp.py,sha256=1awALuAXVpQH64l7vQlM8gvxLDix4c1-6DV3nG5RHu4,3881
@@ -146,8 +146,8 @@ pixeltable/utils/pytorch.py,sha256=6RvOCjy_QV4gc-aht-3d0zoASkuv-warfpl87vgmuKw,3
 pixeltable/utils/s3.py,sha256=huA5hxDGkPIu18zWet76o0FsO7Vbtp-iPmnOzCe-MvA,586
 pixeltable/utils/sql.py,sha256=j_tj0h4ffm-DhUIJbvGphxrVyBKlNTwDKqWGhRQ5_PY,795
 pixeltable/utils/transactional_directory.py,sha256=UGzCrGtLR3hEEf8sYGuWBzLVFAEQml3vdIavigWeTBM,1349
-pixeltable-0.2.23.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-pixeltable-0.2.23.dist-info/METADATA,sha256=WYtMLp2PmlhHkQl6WUyvJFuJ2X-_lSB1VUO3i56dKwc,18112
-pixeltable-0.2.23.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
-pixeltable-0.2.23.dist-info/entry_points.txt,sha256=TNI1Gb5vPwFrTdw6TimSYjO8FeK8c_HuPr28vcf7o_I,108
-pixeltable-0.2.23.dist-info/RECORD,,
+pixeltable-0.2.24.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+pixeltable-0.2.24.dist-info/METADATA,sha256=pOqK9GV-mL4WjzzxeM9w4sJGZsJNSN05DhiTslxJLtk,18112
+pixeltable-0.2.24.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
+pixeltable-0.2.24.dist-info/entry_points.txt,sha256=TNI1Gb5vPwFrTdw6TimSYjO8FeK8c_HuPr28vcf7o_I,108
+pixeltable-0.2.24.dist-info/RECORD,,