com.phronemophobic.whisper
bytes->floatsclj
(bytes->floats bs)default-mono-formatclj
(default-mono-format)downsampleclj
(downsample bs from-sample-rate to-sample-rate)Convert the sound samples, bs from from-sample-rate to to-sample-rate.
The sound samples must be pcmf32.
Convert the sound samples, `bs` from `from-sample-rate` to `to-sample-rate`. The sound samples must be pcmf32.
pcms16->pcmf32clj
(pcms16->pcmf32 bs)Convert a byte array of PCM signed shorts, return a byte array of pcm signed 32 bit floats.
Convert a byte array of PCM signed shorts, return a byte array of pcm signed 32 bit floats.
record-and-transcribeclj
(record-and-transcribe model-path)Starts recording audio from the default system microphone.
Returns a function that will stop recording and return the transcribed text when called.
Starts recording audio from the default system microphone. Returns a function that will stop recording and return the transcribed text when called.
record-audioclj
(record-audio)(record-audio opts)Records audio from the default microphone.
Returns a function that will stop recording and return the recorded audio as a byte array in pcmf32 with a sample rate of 44100.
Records audio from the default microphone. Returns a function that will stop recording and return the recorded audio as a byte array in pcmf32 with a sample rate of 44100.
transcribeclj
(transcribe model-path bs)Given a model-path to a whisper model and a byte array of
sound samples, return a string with the transcribed text.
The sounds samples must be in pcm signed 32 bit float format with a sampling rate of 16,000.
Given a `model-path` to a whisper model and a byte array of sound samples, return a string with the transcribed text. The sounds samples must be in pcm signed 32 bit float format with a sampling rate of 16,000.
cljdoc is a website building & hosting documentation for Clojure/Script libraries
× close