Native R 'torch' Implementation of 'OpenAI' 'Whisper'

---

Documentation for package ‘whisper’ version 0.4.0

• DESCRIPTION file
• Package NEWS
• Package README

Help Pages

.model_sizes	Model Download Utilities
apply_bpe	Apply BPE Merges
apply_timestamp_rules	Apply Timestamp Token Rules
audio_duration	Get Audio Duration
audio_to_mel	Convert Audio to Mel Spectrogram
beam_search_decode	Beam Search Decode
build_byte_decoder	Build Reverse Byte Decoder
byte_to_token	Convert Byte to BPE Token
clean_text	Clean Transcribed Text
compression_ratio	Compression Ratio
compute_stft	Compute STFT Magnitude
compute_word_timestamps	Compute Word-Level Timestamps
copy_if_exists	Copy Weight if Exists
create_decoder	Create Decoder from Config
create_encoder	Create Encoder from Config
create_mel_filterbank_fallback	Create Mel Filterbank (Fallback)
decode_bpe_bytes	Decode BPE Bytes Back to Text
decode_timestamp	Decode Timestamp Token
decode_with_fallback	Decode with Temperature Fallback
detect_language	Detect Language
detect_language_from_mel	Detect Language from Mel Spectrogram
detect_language_from_pipeline	Detect Language from Pipeline
download_tokenizer_files	Download Tokenizer Files from HuggingFace
download_whisper_model	Download Model from HuggingFace
dtw_align	DTW Alignment
ensure_tokenizer_files	Ensure Tokenizer Files are Downloaded
expand_kv_cache	Expand KV Cache for Beam Search
extract_segments	Extract Segments with Timestamps
forced_decode	Forced Decode
get_initial_tokens	Get Initial Decoder Tokens
get_model_path	Get Model Cache Path
get_weights_path	Get Path to Model Weights
greedy_decode	Greedy Decoding
group_into_words	Group Subword Tokens into Words
hz_to_mel	Convert Hz to Mel Scale
is_timestamp_token	Check if Token is Timestamp
list_downloaded_models	List Downloaded Models
list_whisper_models	List Available Models
load_audio	Load and Preprocess Audio
load_decoder_weights	Load Decoder Weights
load_encoder_weights	Load Encoder Weights
load_mel_filterbank	Load Pre-computed Mel Filterbank
load_whisper_model	Load Whisper Model
load_whisper_weights	Load Weights from Safetensors
medfilt1	1D Median Filter
mel_to_hz	Convert Mel Scale to Hz
model_exists	Check if Model is Downloaded
pad_or_trim	Pad or Trim Audio to Fixed Length
parse_device	Parse Device Argument
parse_dtype	Parse Dtype Argument
rearrange_kv_cache	Rearrange KV Cache by Beam Indices
sample_decode	Sample Decode
serve	Serve whisper over HTTP
split_audio	Split Long Audio into Chunks
tokenizer_decode	Decode Token IDs to Text
tokenizer_encode	Encode Text to Token IDs
transcribe	Transcribe Audio
transcribe_chunk	Transcribe Single Chunk
transcribe_long	Transcribe Long Audio
whisper_attention	Multi-Head Self-Attention
whisper_config	Whisper Model Configurations
whisper_decoder	Text Decoder
whisper_decoder_layer	Decoder Layer
whisper_device	Get Default Device
whisper_dtype	Get Default Dtype
whisper_encoder	Audio Encoder
whisper_encoder_layer	Encoder Layer
whisper_language_table	Whisper Language Table
whisper_lang_from_id	Get Language Code from Token ID
whisper_lang_token	Get Language Token ID
whisper_model	Whisper Model Module
whisper_pipeline	Create a Whisper Pipeline
WHISPER_SAMPLE_RATE	Whisper Audio Constants
whisper_special_tokens	Special Token IDs
whisper_tokenizer	Create Whisper Tokenizer
whisper_tune_gc	Tune torch's CUDA garbage collection for whisper inference