Package io.github.givimad.whisperjni

Class WhisperFullParams

java.lang.Object
io.github.givimad.whisperjni.WhisperFullParams
public class WhisperFullParams extends Object
The WhisperFullParams instances needed to configure full whisper execution
Author:
Miguel Álvarez Díez - Initial contribution

  • Field Details

    • nThreads

      public int nThreads
      Number of thread, 0 for max cores

    • audioCtx

      public int audioCtx
      Overwrite the audio context size (0 = use default)

    • nMaxTextCtx

      public int nMaxTextCtx
      Max tokens to use from past text as prompt for the decoder

    • offsetMs

      public int offsetMs
      Start offset in ms

    • durationMs

      public int durationMs
      Audio duration to process in ms

    • translate

      public boolean translate
      Translate

    • noTimestamps

      public boolean noTimestamps
      Do not generate timestamps

    • detectLanguage

      public boolean detectLanguage
      Detect language

    • language

      public String language
      Language

    • initialPrompt

      public String initialPrompt
      Initial prompt

    • noContext

      public boolean noContext
      Do not use past transcription (if any) as initial prompt for the decoder

    • singleSegment

      public boolean singleSegment
      Force single segment output (useful for streaming)

    • printSpecial

      public boolean printSpecial
      Print special tokens

    • printProgress

      public boolean printProgress
      Print progress information

    • printRealtime

      public boolean printRealtime
      Print results from within whisper.cpp (avoid it, use callback instead)

    • printTimestamps

      public boolean printTimestamps
      Print timestamps for each text segment when printing realtime

    • suppressBlank

      public boolean suppressBlank
      Decoder option

    • suppressNonSpeechTokens

      public boolean suppressNonSpeechTokens
      Tokenizer option

    • speedUp

      public boolean speedUp
      Speed-up the audio by 2x using Phase Vocoder

    • temperature

      public float temperature
      Initial decoding temperature

    • maxInitialTs

      public float maxInitialTs
      Refer to library

    • lengthPenalty

      public float lengthPenalty
      Refer to library

    • temperatureInc

      public float temperatureInc
      Refer to library

    • entropyThold

      public float entropyThold
      Refer to library

    • logprobThold

      public float logprobThold
      Refer to library

    • noSpeechThold

      public float noSpeechThold
      Refer to library

    • greedyBestOf

      public int greedyBestOf
      Specific to greedy sampling strategy

    • beamSearchBeamSize

      public int beamSearchBeamSize
      Specific to bean search sampling strategy

    • beamSearchPatience

      public float beamSearchPatience
      Specific to bean search sampling strategy

    • grammar

      public WhisperGrammar grammar
      GBNF grammar.

    • grammarPenalty

      public float grammarPenalty
      Penalty for non grammar tokens.

  • Constructor Details