U2 updates #25
U2 updates #25
Conversation
| @@ -35,18 +36,28 @@ classifier = pipeline( | |||
| ) | |||
| ``` | |||
|
|
|||
| This pipeline expects the audio data as a NumPy array. All the preprocessing of the raw audio data will be conveniently | |||
| handled for us by the pipeline. Let's pick an example to try it out: | |||
| All the preprocessing of the raw audio data will be conveniently handled for us by the pipeline, including any resampling. | |||
There was a problem hiding this comment.
(Once huggingface/transformers#23445 is merged)
|
|
||
| minds = load_dataset("PolyAI/minds14", name="en-AU", split="train") | ||
| minds = minds.cast_column("audio", Audio(sampling_rate=16_000)) |
There was a problem hiding this comment.
Think it's cleaner if we don't have to do any data pre-/post-processing and let the pipeline handle this
There was a problem hiding this comment.
Same as above, IMO it is good to reinforce the idea of sampling rates.
|
|
||
| ```py | ||
| example = minds[0] | ||
| example["transcription"] | ||
| "ich möchte gerne Geld auf mein Konto einzahlen" | ||
| ``` | ||
|
|
||
| Find a pre-trained ASR model for German language on the 🤗 Hub, instantiate a pipeline, and transcribe the example: | ||
| Next, we can find a pre-trained ASR model for German language on the 🤗 Hub, instantiate a pipeline, and transcribe the example. | ||
| Here, we'll use the checkpoint [maxidl/wav2vec2-large-xlsr-german](https://huggingface.co/maxidl/wav2vec2-large-xlsr-german): |
There was a problem hiding this comment.
I personally like having links to the checkpoints on the Hub so that I can look at the model cards
There was a problem hiding this comment.
Lovely, how about we swap the community checkpoint to an official DE checkpoint for XLSR: https://huggingface.co/facebook/wav2vec2-large-xlsr-53-german
| the right format for a model | ||
| - if the result isn't ideal, this still gives you a quick baseline for future fine-tuning | ||
| - once you fine-tune a model on your custom data and share it on Hub, the whole community will be able to use it quickly | ||
| and effortlessly via the `pipeline()` method making AI more accessible. | ||
|
|
||
| and effortlessly via the `pipeline()` method, making AI more accessible |
There was a problem hiding this comment.
Consistency with previous bullet points
|
|
||
| minds = load_dataset("PolyAI/minds14", name="de-DE", split="train") | ||
| minds = minds.cast_column("audio", Audio(sampling_rate=16_000)) |
There was a problem hiding this comment.
Personal preference: It makes sense to explicitly resample here, just to reinforce the idea of sampling rates to the attendee.
We can later on explicitly write that sampling_rate handles different rates automagically.
WDYT?
|
|
||
| ```py | ||
| example = minds[0] | ||
| example["transcription"] | ||
| "ich möchte gerne Geld auf mein Konto einzahlen" | ||
| ``` | ||
|
|
||
| Find a pre-trained ASR model for German language on the 🤗 Hub, instantiate a pipeline, and transcribe the example: | ||
| Next, we can find a pre-trained ASR model for German language on the 🤗 Hub, instantiate a pipeline, and transcribe the example. | ||
| Here, we'll use the checkpoint [maxidl/wav2vec2-large-xlsr-german](https://huggingface.co/maxidl/wav2vec2-large-xlsr-german): |
There was a problem hiding this comment.
Lovely, how about we swap the community checkpoint to an official DE checkpoint for XLSR: https://huggingface.co/facebook/wav2vec2-large-xlsr-53-german
|
|
||
| minds = load_dataset("PolyAI/minds14", name="en-AU", split="train") | ||
| minds = minds.cast_column("audio", Audio(sampling_rate=16_000)) |
There was a problem hiding this comment.
Same as above, IMO it is good to reinforce the idea of sampling rates.
Small updates to U2 (some formatting, some updating of the code samples)