NOTE: upstream not implemented, ref: https://arxiv.org/pdf/2204.05424.pdf

ref: https://github.com/openai/whisper/blob/f82bc59f5ea234d4b97fb2860842ed38519f7e65/whisper/transcribe.py#L264

Return all module attributes. Lazy load the module on first access.

Lazy load the module on first access.

Create a lazy loaded module. Args: local_name (str): The name of the module in the parent's globals. parent_module_gl

Empty init method. This will raise a ``RuntimeError``.

Returns a list of available audio devices on the system.

Download a preconverted model from a given model name. Currently it doesn't support custom preconverted models yet. PRs are welcome. Args:

Run the Whisper encoder on the log mel spectrogram stored inside the provided whisper context. Make sure to call whisper_pcm_to_mel() or whisper_set_m

Some special tokens

Get token data for the specified token in the specified segment. This contains probabilities, timestamps, etc.

Get the probability of the specified token in the specified segment.

Get the token text of the specified token in the specified segment.

Language id associated with the context's default state

Token logits obtained from the last call to whisper_decode() The logits for the last token are stored in the last row Rows: n_tokens Cols: n_vocab

language functions. Returns a vector of probabilities for each language.

Returns short string of specified language id, raise exception if nullptr is returned

Returns largest language id

Get mel spectrogram length

Get number of text context

Convert RAW PCM audio to log mel spectrogram. The resulting spectrogram is stored inside the provided whisper context. Returns 0 on success. This is t

Set the callback for starting the encoder. Do not use this function unless you know what you are doing. Defaults to None.

Set the user data to be passed to the encoder begin callback. Defaults to None. See set_encoder_begin_callback.

Set the callback for each decoder to filter obtained logits. Do not use this function unless you know what you are doing. Defaults to None.

Set the user data to be passed to the logits filter callback. Defaults to None. See set_logits_filter_callback.

Low-level API for setting custom log mel spectrogram. The resulting spectrogram is stored inside the provided whisper context.

Set tokens to provide the model as initial input. These tokens are prepended to any existing text content from a previous call. Calling this more than

Streaming transcription from microphone. Note that this function is blocking. Args: length_ms (int, optional): Length of audio to

Run the Whisper decoder to obtain the logits and probabilities for the next token. Make sure to call whisper_encode() first. tokens + n_tokens is the

[EXPERIMENTAL] max segment length in characters defaults to 0 (no limit)