Error in embed_audio.ipynb

[jongalon]

I was trying to use the embed_audio.ipynb notebook to learn how to implement perch for extracting embeddings. However. I've got an error in the forth cell.

---

TypeError Traceback (most recent call last)
Cell In[11], line 13
10 config.source_file_patterns = ['gs://chirp-public-bucket/soundscapes/powdermill/Recording*/*.wav'] #@param
11 config.output_dir = embeddings_path.as_posix()
---> 13 model_key, embedding_dim, model_config = model_configs.get_preset_model_config(model_choice)
14 config.embed_fn_config.model_key = model_key
15 config.embed_fn_config.model_config = model_config

TypeError: cannot unpack non-iterable PresetInfo object

---

So, I had to change the following code:

model_key, embedding_dim, model_config = model_configs.get_preset_model_config(model_choice)
config.embed_fn_config.model_key = model_key
config.embed_fn_config.model_config = model_config

for this one:

preset_info = model_configs.get_preset_model_config(model_choice)
config.embed_fn_config.model_key = preset_info.model_key
config.embed_fn_config.model_config = preset_info.model_config

and it worked.

So, I wanted to know if there is something wrong with the code or if I got a bad installation of the repository. I am trying to use it in a studio in lightnin.ai.

Thanks!!

[sdenton4]

Hello!
Sorry for the slow response - you are correct, and we'll have your fix checked in shortly.

Note that we have a new version of the agile modeling flow available here:

• Hoplite repo: https://github.com/google-research/perch-hoplite
• Embed notebook: https://github.com/google-research/perch-hoplite/blob/main/perch_hoplite/agile/1_embed_audio_v2.ipynb
• Classifier training notebook: https://github.com/google-research/perch-hoplite/blob/main/perch_hoplite/agile/2_agile_modeling_v2.ipynb

The new repo simplifies dependency management, provides large-scale approximate nearest neighbor search, and improves on our classifier training methodology. Hope it's helpful!