mirror of
https://github.com/ggml-org/whisper.cpp.git
synced 2026-10-21 09:41:29 +02:00
ruby : transcribe without GVL, accept more MemoryViews, Windows support, fix memory size report, improve document (#3775)
* Change MemoryView example using NDAV * Add note on audio attributes for #full and #full_parallel * Support more variants of MemoryView * Use IO.popen instead of Kernel.` for Windows compatibility * Use cmake's -C option instead of multiple -D options * Fix memsize calculation * Remove unused argument * Add is_interrupted field to abort callback container * Fix RBS syntax * Address document comment for RDoc * Add .document for RDoc * Add .rdoc_options * Run #full without GVL * Initialize callbacks with nil * Specify implicity Whisper::Params to distinguish from Whisper::Context::Params * Run callbacks without GVL * Call log callback with GVL * Run full_parallel without GVL * Run transcribe without GVL * Fix ruby_whisper_lock_gvl and ruby_whisper_unlock_gvl * Fix return value of encoder_begin_callback * Report GVL unlocking from transcribe * Remove unused interface * Restore overload of full_parallel * Close process * Fix struct name * Make is_without_gvl thread local * Use rb_thread_call_with_gvl instead of global variable * Retrieve instance variable in GVL * Narrow acceptable MemoryView format * Fix option cache path * Reduce files in package * Use append_cflags * Add ext/*.rb to task dependencies * Use copy instead of cp * Make TestPackage more portable * Patch for lower version Ruby * Make build scripts more portable * Add Windows support * Don't raise exceptions
This commit is contained in:
KITAITI Makoto
@@ -5,10 +5,10 @@ module Whisper
|
||||
end
|
||||
|
||||
type log_callback = ^(Integer level, String message, Object user_data) -> void
|
||||
type new_segment_callback = ^(Whisper::Context, void, Integer n_new, Object user_data) -> void
|
||||
type progress_callback = ^(Whisper::Context, void, Integer progress, Object user_data) -> void
|
||||
type encoder_begin_callback = ^(Whisper::Context, void, Object user_data) -> void
|
||||
type abort_callback = ^(Whisper::Context, void, Object user_data) -> boolish
|
||||
type new_segment_callback = ^(Whisper::Context, untyped, Integer n_new, Object user_data) -> void
|
||||
type progress_callback = ^(Whisper::Context, untyped, Integer progress, Object user_data) -> void
|
||||
type encoder_begin_callback = ^(Whisper::Context, untyped, Object user_data) -> void
|
||||
type abort_callback = ^(Whisper::Context, untyped, Object user_data) -> boolish
|
||||
|
||||
VERSION: String
|
||||
LOG_LEVEL_NONE: Integer
|
||||
@@ -52,11 +52,11 @@ module Whisper
|
||||
# puts text
|
||||
# end
|
||||
#
|
||||
# If n_processors is greater than 1, you cannot set any callbacks including
|
||||
# If `n_processors` is greater than 1, you cannot set any callbacks including
|
||||
# new_segment_callback, progress_callback, encoder_begin_callback, abort_callback,
|
||||
# and log_callback set by Whisper.log_set
|
||||
def transcribe: (path, Params, ?n_processors: Integer) -> self
|
||||
| (path, Params, ?n_processors: Integer) { (String) -> void } -> self
|
||||
def transcribe: (path, Whisper::Params, ?n_processors: Integer) -> self
|
||||
| (path, Whisper::Params, ?n_processors: Integer) { (String) -> void } -> self
|
||||
|
||||
def model_n_vocab: () -> Integer
|
||||
def model_n_audio_ctx: () -> Integer
|
||||
@@ -74,7 +74,7 @@ module Whisper
|
||||
# puts segment.text
|
||||
# end
|
||||
#
|
||||
# Returns an Enumerator if no block given:
|
||||
# Returns an `Enumerator` if no block given:
|
||||
#
|
||||
# whisper.transcribe("path/to/audio.wav", params)
|
||||
# enum = whisper.each_segment
|
||||
@@ -91,25 +91,25 @@ module Whisper
|
||||
#
|
||||
def full_lang_id: () -> Integer
|
||||
|
||||
# Start time of a segment indexed by +segment_index+ in centiseconds (10 times milliseconds).
|
||||
# Start time of a segment indexed by `segment_index` in centiseconds (10 times milliseconds).
|
||||
#
|
||||
# full_get_segment_t0(3) # => 1668 (16680 ms)
|
||||
#
|
||||
def full_get_segment_t0: (Integer) -> Integer
|
||||
|
||||
# End time of a segment indexed by +segment_index+ in centiseconds (10 times milliseconds).
|
||||
# End time of a segment indexed by `segment_index` in centiseconds (10 times milliseconds).
|
||||
#
|
||||
# full_get_segment_t1(3) # => 1668 (16680 ms)
|
||||
#
|
||||
def full_get_segment_t1: (Integer) -> Integer
|
||||
|
||||
# Whether the next segment indexed by +segment_index+ is predicated as a speaker turn.
|
||||
# Whether the next segment indexed by `segment_index` is predicated as a speaker turn.
|
||||
#
|
||||
# full_get_segment_speacker_turn_next(3) # => true
|
||||
#
|
||||
def full_get_segment_speaker_turn_next: (Integer) -> (true | false)
|
||||
|
||||
# Text of a segment indexed by +segment_index+.
|
||||
# Text of a segment indexed by `segment_index`.
|
||||
#
|
||||
# full_get_segment_text(3) # => "ask not what your country can do for you, ..."
|
||||
#
|
||||
@@ -117,27 +117,27 @@ module Whisper
|
||||
|
||||
def full_get_segment_no_speech_prob: (Integer) -> Float
|
||||
|
||||
# Run the entire model: PCM -> log mel spectrogram -> encoder -> decoder -> text
|
||||
# Not thread safe for same context
|
||||
# Run the entire model: PCM -> log mel spectrogram -> encoder -> decoder -> text
|
||||
# Not thread safe for same context
|
||||
# Uses the specified decoding strategy to obtain the text.
|
||||
#
|
||||
# The second argument +samples+ must be an array of samples, respond to :length, or be a MemoryView of an array of float. It must be 32 bit float PCM audio data.
|
||||
# The second argument `samples` must be an array of samples, respond to `:length`, or be a MemoryView of an array of float. It must be 32 bit float PCM audio data.
|
||||
#
|
||||
def full: (Params, Array[Float] samples, ?Integer n_samples) -> self
|
||||
| (Params, _Samples, ?Integer n_samples) -> self
|
||||
def full: (Whisper::Params, Array[Float] samples, ?Integer n_samples) -> self
|
||||
| (Whisper::Params, _Samples, ?Integer n_samples) -> self
|
||||
|
||||
# Split the input audio in chunks and process each chunk separately using whisper_full_with_state()
|
||||
# Result is stored in the default state of the context
|
||||
# Not thread safe if executed in parallel on the same context.
|
||||
# It seems this approach can offer some speedup in some cases.
|
||||
# Split the input audio in chunks and process each chunk separately using `whisper_full_with_state()`
|
||||
# Result is stored in the default state of the context
|
||||
# Not thread safe if executed in parallel on the same context.
|
||||
# It seems this approach can offer some speedup in some cases.
|
||||
# However, the transcription accuracy can be worse at the beginning and end of each chunk.
|
||||
#
|
||||
# If n_processors is greater than 1, you cannot set any callbacks including
|
||||
# If `n_processors` is greater than 1, you cannot set any callbacks including
|
||||
# new_segment_callback, progress_callback, encoder_begin_callback, abort_callback,
|
||||
# and log_callback set by Whisper.log_set
|
||||
def full_parallel: (Params, Array[Float], ?Integer n_samples) -> self
|
||||
| (Params, _Samples, ?Integer n_samples) -> self
|
||||
| (Params, _Samples, ?Integer? n_samples, Integer n_processors) -> self
|
||||
def full_parallel: (Whisper::Params, Array[Float], ?Integer n_samples) -> self
|
||||
| (Whisper::Params, _Samples, ?Integer n_samples) -> self
|
||||
| (Whisper::Params, _Samples, ?Integer? n_samples, Integer n_processors) -> self
|
||||
|
||||
def to_srt: () -> String
|
||||
def to_webvtt: () -> String
|
||||
@@ -217,35 +217,35 @@ module Whisper
|
||||
def translate: () -> (true | false)
|
||||
def no_context=: (boolish) -> boolish
|
||||
|
||||
# If true, does not use past transcription (if any) as initial prompt for the decoder.
|
||||
# If `true`, does not use past transcription (if any) as initial prompt for the decoder.
|
||||
#
|
||||
def no_context: () -> (true | false)
|
||||
|
||||
def single_segment=: (boolish) -> boolish
|
||||
|
||||
# If true, forces single segment output (useful for streaming).
|
||||
# If `true`, forces single segment output (useful for streaming).
|
||||
#
|
||||
def single_segment: () -> (true | false)
|
||||
|
||||
def print_special=: (boolish) -> boolish
|
||||
|
||||
# If true, prints special tokens (e.g. <SOT>, <EOT>, <BEG>, etc.).
|
||||
# If `true`, prints special tokens (e.g. <SOT>, <EOT>, <BEG>, etc.).
|
||||
#
|
||||
def print_special: () -> (true | false)
|
||||
|
||||
def print_progress=: (boolish) -> boolish
|
||||
|
||||
# If true, prints progress information.
|
||||
# If `true`, prints progress information.
|
||||
#
|
||||
def print_progress: () -> (true | false)
|
||||
|
||||
def print_realtime=: (boolish) -> boolish
|
||||
|
||||
# If true, prints results from within whisper.cpp. (avoid it, use callback instead)
|
||||
# If `true`, prints results from within whisper.cpp. (avoid it, use callback instead)
|
||||
#
|
||||
def print_realtime: () -> (true | false)
|
||||
|
||||
# If true, prints timestamps for each text segment when printing realtime.
|
||||
# If `true`, prints timestamps for each text segment when printing realtime.
|
||||
#
|
||||
def print_timestamps=: (boolish) -> boolish
|
||||
|
||||
@@ -253,19 +253,19 @@ module Whisper
|
||||
|
||||
def suppress_blank=: (boolish) -> boolish
|
||||
|
||||
# If true, suppresses blank outputs.
|
||||
# If `true`, suppresses blank outputs.
|
||||
#
|
||||
def suppress_blank: () -> (true | false)
|
||||
|
||||
def suppress_nst=: (boolish) -> boolish
|
||||
|
||||
# If true, suppresses non-speech-tokens.
|
||||
# If `true`, suppresses non-speech-tokens.
|
||||
#
|
||||
def suppress_nst: () -> (true | false)
|
||||
|
||||
def token_timestamps=: (boolish) -> boolish
|
||||
|
||||
# If true, enables token-level timestamps.
|
||||
# If `true`, enables token-level timestamps.
|
||||
#
|
||||
def token_timestamps: () -> (true | false)
|
||||
|
||||
@@ -277,16 +277,16 @@ module Whisper
|
||||
|
||||
def split_on_word=: (boolish) -> boolish
|
||||
|
||||
# If true, split on word rather than on token (when used with max_len).
|
||||
# If `true`, split on word rather than on token (when used with max_len).
|
||||
#
|
||||
def split_on_word: () -> (true | false)
|
||||
|
||||
def initial_prompt=: (_ToS) -> _ToS
|
||||
def carry_initial_prompt=: (boolish) -> boolish
|
||||
|
||||
# Tokens to provide to the whisper decoder as initial prompt
|
||||
# these are prepended to any existing text context from a previous call
|
||||
# use whisper_tokenize() to convert text to tokens.
|
||||
# Tokens to provide to the whisper decoder as initial prompt
|
||||
# these are prepended to any existing text context from a previous call
|
||||
# use whisper_tokenize() to convert text to tokens.
|
||||
# Maximum of whisper_n_text_ctx()/2 tokens are used (typically 224).
|
||||
#
|
||||
def initial_prompt: () -> (String | nil)
|
||||
@@ -294,7 +294,7 @@ module Whisper
|
||||
|
||||
def diarize=: (boolish) -> boolish
|
||||
|
||||
# If true, enables diarization.
|
||||
# If `true`, enables diarization.
|
||||
#
|
||||
def diarize: () -> (true | false)
|
||||
|
||||
@@ -423,7 +423,7 @@ module Whisper
|
||||
#
|
||||
def on_new_segment: { (Segment) -> void } -> void
|
||||
|
||||
# Hook called on progress update. Yields each progress Integer between 0 and 100.
|
||||
# Hook called on progress update. Yields each progress `Integer` between 0 and 100.
|
||||
#
|
||||
def on_progress: { (Integer progress) -> void } -> void
|
||||
|
||||
@@ -431,7 +431,7 @@ module Whisper
|
||||
#
|
||||
def on_encoder_begin: { () -> void } -> void
|
||||
|
||||
# Call block to determine whether abort or not. Return +true+ when you want to abort.
|
||||
# Call block to determine whether abort or not. Return `true` when you want to abort.
|
||||
#
|
||||
# params.abort_on do
|
||||
# if some_condition
|
||||
@@ -504,13 +504,13 @@ module Whisper
|
||||
|
||||
# Yields each Whisper::Token:
|
||||
#
|
||||
# whisper.each_segment.first.each_token do |token|
|
||||
# p token
|
||||
# end
|
||||
# whisper.each_segment.first.each_token do |token|
|
||||
# p token
|
||||
# end
|
||||
#
|
||||
# Returns an Enumerator if no block is given:
|
||||
# Returns an `Enumerator` if no block is given:
|
||||
#
|
||||
# whisper.each_segment.first.each_token.to_a # => [#<Whisper::Token>, ...]
|
||||
# whisper.each_segment.first.each_token.to_a # => [#<Whisper::Token>, ...]
|
||||
#
|
||||
def each_token: { (Token) -> void } -> void
|
||||
| () -> Enumerator[Token]
|
||||
@@ -518,7 +518,7 @@ module Whisper
|
||||
def to_webvtt_cue: () -> String
|
||||
|
||||
|
||||
# Possible keys: :start_time, :end_time, :text, :no_speech_prob, :speaker_turn_next
|
||||
# Possible keys: `:start_time`, `:end_time`, `:text`, `:no_speech_prob`, `:speaker_turn_next`
|
||||
#
|
||||
# whisper.each_segment do |segment|
|
||||
# segment => {start_time:, end_time:, text:, no_speech_prob:, speaker_turn_next:}
|
||||
@@ -569,7 +569,7 @@ module Whisper
|
||||
|
||||
# [EXPERIMENTAL] Token-level timestamps with DTW
|
||||
#
|
||||
# Do not use if you haven't computed token-level timestamps with dtw.
|
||||
# Do not use if you haven't computed token-level timestamps with dtw.
|
||||
# Roughly corresponds to the moment in audio in which the token was output.
|
||||
#
|
||||
def t_dtw: () -> Integer
|
||||
@@ -580,14 +580,14 @@ module Whisper
|
||||
|
||||
# Start time of the token.
|
||||
#
|
||||
# Token-level timestamp data.
|
||||
# Token-level timestamp data.
|
||||
# Do not use if you haven't computed token-level timestamps.
|
||||
#
|
||||
def start_time: () -> Integer
|
||||
|
||||
# End time of the token.
|
||||
#
|
||||
# Token-level timestamp data.
|
||||
# Token-level timestamp data.
|
||||
# Do not use if you haven't computed token-level timestamps.
|
||||
#
|
||||
def end_time: () -> Integer
|
||||
|
||||
Reference in New Issue
Block a user