it- dZddlmZddlmZddlmZmZddlZddl Z ddl Z ddl mZmZmZmZddlmZmZe dded eed eeee j.ffd Zej2 dd e j.ded eed eeee j.ffd Zej2 dd eded eed eefdZGddZGddZy)z # Signal processing )singledispatch) zip_longest)OptionalUnionN) AnnotationSegmentSlidingWindowFeatureTimeline)pairwisestring_generatoronsetoffset initial_statectd)a(Batch) hysteresis thresholding Parameters ---------- scores : numpy.ndarray or SlidingWindowFeature (num_chunks, num_frames, num_classes)- or (num_frames, num_classes)-shaped scores. onset : float, optional Onset threshold. Defaults to 0.5. offset : float, optional Offset threshold. Defaults to `onset`. initial_state : np.ndarray or bool, optional Initial state. Returns ------- binarized : same as scores Binarized scores with same shape and type as scores. Reference --------- https://stackoverflow.com/questions/23289976/how-to-find-zero-crossings-with-hysteresis z=scores must be of type numpy.ndarray or SlidingWindowFeatures)NotImplementedError)scoresr rrs p/Volumes/fast/ai/experiments/voice-extract-mac/.venv/lib/python3.12/site-packages/pyannote/audio/utils/signal.pybinarizer,s: G rc |xs|}|j\}}tj|}||dddfd||zzk\}nqt|tr |tj |ftz}nAt|tj r'|j|fk(sJ|jtk(sJtj||dfj}||kD}||k|z}tjtt|Dcgc]}tj|dc}ddij} | jstj|t|zStj |d} tjtj"||dfj} tj$| || | | | dz ff|Scc}w) a(Batch) hysteresis thresholding Parameters ---------- scores : numpy.ndarray (num_frames, num_classes)-shaped scores. onset : float, optional Onset threshold. Defaults to 0.5. offset : float, optional Offset threshold. Defaults to `onset`. initial_state : np.ndarray or bool, optional Initial state. Returns ------- binarized : same as scores Binarized scores with same shape and type as scores. Nr?)dtype fillvalue)axis)shapenp nan_to_num isinstanceboolonesndarrayrtileTarraylistrnonzerosize zeros_likecumsumarangewhere) rr rr batch_size num_frameson off_or_onoonwell_defined_idxsame_assampless rbinarize_ndarrayr6Ns4_uF#\\J ]]6 "Fq!t uv~(>> M4 (% T(JJ M2:: .""zm333""d***GGMJ?;==M %B&B&Ixx [C#2::c?1-C Rr RSa  }}V40=@@ii *Gggbii +j!_=??G 88G-gw{.BCCDm Ds=Gc z|xs|}|jjdk(r{|jj\}}tj|jd||}t ||||}t dtj|d||z|jS|jjdk(r~|jj\}}}tj|jd||| }t ||||}t dtj|d ||| z|jStd ) a(Batch) hysteresis thresholding Parameters ---------- scores : SlidingWindowFeature (num_chunks, num_frames, num_classes)- or (num_frames, num_classes)-shaped scores. onset : float, optional Onset threshold. Defaults to 0.5. offset : float, optional Offset threshold. Defaults to `onset`. initial_state : np.ndarray or bool, optional Initial state. Returns ------- binarized : same as scores Binarized scores with same shape and type as scores. z f k -> k f)fk)r rr?z k f -> f kzc f k -> (c k) f)cr9r:z(c k) f -> c f kz[Shape of scores must be (num_chunks, num_frames, num_classes) or (num_frames, num_classes).) datandimreinops rearrangerr sliding_window ValueError) rr rrr/ num_classesr> binarized num_chunkss r binarize_swfrGsB6_uF {{1"(++"3"3 K \Z;W fM $ y,* T U  ! !    Q .4kk.?.?+ J  KK+zZ;  fM $ -z[   ! !    i  rc `eZdZdZ d dedeededededef fd Zd ed efd Z xZ S) BinarizeaBinarize detection scores using hysteresis thresholding Parameters ---------- onset : float, optional Onset threshold. Defaults to 0.5. offset : float, optional Offset threshold. Defaults to `onset`. min_duration_on : float, optional Remove active regions shorter than that many seconds. Defaults to 0s. min_duration_off : float, optional Fill inactive regions shorter than that many seconds. Defaults to 0s. pad_onset : float, optional Extend active regions by moving their start time by that many seconds. Defaults to 0s. pad_offset : float, optional Extend active regions by moving their end time by that many seconds. Defaults to 0s. Reference --------- Gregory Gelly and Jean-Luc Gauvain. "Minimum Word Error Training of RNN-based Voice Activity Detection", InterSpeech 2015. r rmin_duration_onmin_duration_off pad_onset pad_offsetc~t|||_|xs||_||_||_||_||_yN)super__init__r rrLrMrJrK)selfr rrJrKrLrM __class__s rrQzBinarize.__init__s@  o "$. 0rrreturnc |jj\}}|j}t|Dcgc]}||j}}t }t }t|jjD]\} } |j| n|j| } t|} |d} | d|jkD}t|dd| ddD]]\}}|rB||jkst| |jz ||j z}| ||| f<|} d}J||jkDsZ|} d}_|st| |jz |j z}| ||| f<|j dkDs|jdkDs|j"dkDr|j%|j"}|j&dkDr@t)|j+D]$\}} |j,|j&ks ||| f=&|Scc}w)zBinarize detection scores Parameters ---------- scores : SlidingWindowFeature Detection scores. Returns ------- active : Annotation Binarized scores. NrrFT)collar)r>rrBrangemiddlerr enumerater%labelsnextr ziprrrLrMrKsupportrJr' itertracksduration)rRrr/rDframesi timestampsactivetrack_generatorr:k_scoreslabeltrackstart is_activetyregionsegments r__call__zBinarize.__call__s#)++"3"3 K&&05j0AB0A1fQi&&0A B*,$V[[]]3KAx.AFMM!4DE)EqME  djj0IJqrNHQRL914;;!()?T__AT!U05vu}- !$) 4::~ !$( !:& !7T__9LM(-vu}%;4B ??S DNNS$8DeZdZdZ ddedeffd ZdefdZxZS)PeakzPeak detection Parameters ---------- alpha : float, optional Peak threshold. Defaults to 0.5 min_duration : float, optional Minimum elapsed time between two consecutive peaks. Defaults to 1 second. alpha min_durationcFtt| ||_||_yrO)rPrxrQryrz)rRryrzrSs rrQz Peak.__init__Ls! dD"$ (rrc |jdk7r tdt|}|j}|j}t dt tj|j|z }tjj|dd|d}tj|Dcgc]$}|||jkDs||j&c}}tj |dj"g|||j$gg} t'} t)t+| D]%\}\} } t-| | } | j/| '| Scc}w)zPeak detection Parameter --------- scores : SlidingWindowFeature Detection scores. Returns ------- segmentation : Timeline Partition. rz$Peak expects one-dimensional scores.N)orderr) dimensionrClenrBstepmaxintrrintrzscipysignal argrelmaxr&ryrYhstackriendr rZr radd)rRrr/ra precisionr}indicesrb peak_time boundaries segmentationrirrns rroz Peak.__call__Us6   q CD D[ &&KK As2774#4#4y#@ABC,,((%(@CHH'. Iw!&)djj2HVAY  w I YY 19vj?Q?U?U>VWX z (*)=>OA|seS)G   W % ? Js -EE)rr;) rprqrrrsrtrQr rorurvs@rrxrxAs4!)))"3"rrx)rNN)rs functoolsr itertoolsrtypingrrr@numpyr scipy.signalr pyannote.corerrr r pyannote.core.utils.generatorsr r rtr!r#rregisterr6rGrIrxrrrsH:%!" MME"7;   UOE$ "234 B "7; > JJ> > UO>E$ "234 >>B "$( ; ; ;  UO; D> ; ; |ood66r