Function deepspeech_sys::DS_AudioToInputVector
source · pub unsafe extern "C" fn DS_AudioToInputVector(
aBuffer: *const c_short,
aBufferSize: c_uint,
aSampleRate: c_uint,
aNCep: c_uint,
aNContext: c_uint,
aMfcc: *mut *mut f32,
aNFrames: *mut c_int,
aFrameLen: *mut c_int
)
Expand description
@brief Given audio, return a vector suitable for input to a DeepSpeech model trained with the given parameters.
Extracts MFCC features from a given audio signal and adds the appropriate amount of context to run inference on a DeepSpeech model trained with the given parameters.
@param aBuffer A 16-bit, mono raw audio signal at the appropriate sample rate. @param aBufferSize The sample-length of the audio signal. @param aSampleRate The sample-rate of the audio signal. @param aNCep The number of cepstrum. @param aNContext The size of the context window. @param[out] aMfcc An array containing features, of shape (@p aNFrames, ncep * ncontext). The user is responsible for freeing the array. @param[out] aNFrames (optional) The number of frames in @p aMfcc. @param[out] aFrameLen (optional) The length of each frame (ncep * ncontext) in @p aMfcc.