Audio Utils#

namespace trt_edgellm

namespace rt

namespace audioUtils

struct AudioData#

#include <audioUtils.h>

Audio data container For input: provide melSpectrogramPath for pre-computed Mel-spectrogram For output: contains generated audio waveform data, sampleRate, and numChannels

Public Members

std::string melSpectrogramPath#: Path to pre-computed Mel-spectrogram file (.npy or .raw)

std::string melSpectrogramFormat#: Format of the mel-spectrogram file: “npy” or “raw”.

std::shared_ptr<Tensor> waveform#: Waveform samples [1, numSamples], FP16, range [-1, 1], CPU.

int32_t sampleRate = {24000}#: Sample rate in Hz.

int32_t numChannels = {1}#: Number of audio channels (typically 1 for mono)

std::vector<std::vector<int32_t>> codebookCodes#: RVQ codebook codes [numCodebooks][seqLen].

bool hasWaveform = {false}#: True if waveform contains valid data.