com.johnsnowlabs.nlp.annotators.audio.HubertForCTC.scala Maven / Gradle / Ivy
The newest version!
/*
* Copyright 2017-2022 John Snow Labs
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package com.johnsnowlabs.nlp.annotators.audio
import com.johnsnowlabs.ml.onnx.{OnnxWrapper, ReadOnnxModel}
import com.johnsnowlabs.ml.tensorflow.{ReadTensorflowModel, TensorflowWrapper}
import com.johnsnowlabs.ml.util.LoadExternalModel.{
detectEngine,
loadJsonStringAsset,
modelSanityCheck,
notSupportedEngineError
}
import com.johnsnowlabs.ml.util.{ONNX, TensorFlow}
import com.johnsnowlabs.nlp._
import com.johnsnowlabs.nlp.annotators.audio.feature_extractor.Preprocessor
import org.apache.spark.ml.util.Identifiable
import org.apache.spark.sql.SparkSession
import org.json4s._
import org.json4s.jackson.JsonMethods._
/** Hubert Model with a language modeling head on top for Connectionist Temporal Classification
* (CTC). Hubert was proposed in HuBERT: Self-Supervised Speech Representation Learning by Masked
* Prediction of Hidden Units by Wei-Ning Hsu, Benjamin Bolte, Yao-Hung Hubert Tsai, Kushal
* Lakhotia, Ruslan Salakhutdinov, Abdelrahman Mohamed.
*
* The annotator takes audio files and transcribes it as text. The audio needs to be provided
* pre-processed an array of floats.
*
* Note that this annotator is currently not supported on Apple Silicon processors such as the
* M1/M2 (Apple Silicon). This is due to the processor not supporting instructions for XLA.
*
* Pretrained models can be loaded with `pretrained` of the companion object:
* {{{
* val speechToText = HubertForCTC.pretrained()
* .setInputCols("audio_assembler")
* .setOutputCol("text")
* }}}
* The default model is `"asr_hubert_large_ls960"`, if no name is provided.
*
* For available pretrained models please see the [[https://sparknlp.org/models Models Hub]].
*
* To see which models are compatible and how to import them see
* [[https://github.com/JohnSnowLabs/spark-nlp/discussions/5669]] and to see more extended
* examples, see
* [[https://github.com/JohnSnowLabs/spark-nlp/blob/master/src/test/scala/com/johnsnowlabs/nlp/annotators/audio/HubertForCTCTestSpec.scala HubertForCTCTestSpec]].
*
* '''References:'''
*
* [[https://arxiv.org/abs/2106.07447 HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units]]
*
* '''Paper Abstract:'''
*
* ''Self-supervised approaches for speech representation learning are challenged by three unique
* problems: (1) there are multiple sound units in each input utterance, (2) there is no lexicon
* of input sound units during the pre-training phase, and (3) sound units have variable lengths
* with no explicit segmentation. To deal with these three problems, we propose the Hidden-Unit
* BERT (HuBERT) approach for self-supervised speech representation learning, which utilizes an
* offline clustering step to provide aligned target labels for a BERT-like prediction loss. A
* key ingredient of our approach is applying the prediction loss over the masked regions only,
* which forces the model to learn a combined acoustic and language model over the continuous
* inputs. HuBERT relies primarily on the consistency of the unsupervised clustering step rather
* than the intrinsic quality of the assigned cluster labels. Starting with a simple k-means
* teacher of 100 clusters, and using two iterations of clustering, the HuBERT model either
* matches or improves upon the state-of-the-art wav2vec 2.0 performance on the Librispeech
* (960h) and Libri-light (60,000h) benchmarks with 10min, 1h, 10h, 100h, and 960h fine-tuning
* subsets. Using a 1B parameter model, HuBERT shows up to 19% and 13% relative WER reduction on
* the more challenging dev-other and test-other evaluation subsets.''
*
* ==Example==
* {{{
* import spark.implicits._
* import com.johnsnowlabs.nlp.base._
* import com.johnsnowlabs.nlp.annotators._
* import com.johnsnowlabs.nlp.annotators.audio.HubertForCTC
* import org.apache.spark.ml.Pipeline
*
* val audioAssembler: AudioAssembler = new AudioAssembler()
* .setInputCol("audio_content")
* .setOutputCol("audio_assembler")
*
* val speechToText: HubertForCTC = HubertForCTC
* .pretrained()
* .setInputCols("audio_assembler")
* .setOutputCol("text")
*
* val pipeline: Pipeline = new Pipeline().setStages(Array(audioAssembler, speechToText))
*
* val bufferedSource =
* scala.io.Source.fromFile("src/test/resources/audio/csv/audio_floats.csv")
*
* val rawFloats = bufferedSource
* .getLines()
* .map(_.split(",").head.trim.toFloat)
* .toArray
* bufferedSource.close
*
* val processedAudioFloats = Seq(rawFloats).toDF("audio_content")
*
* val result = pipeline.fit(processedAudioFloats).transform(processedAudioFloats)
* result.select("text.result").show(truncate = false)
* +------------------------------------------------------------------------------------------+
* |result |
* +------------------------------------------------------------------------------------------+
* |[MISTER QUILTER IS THE APOSTLE OF THE MIDLE CLASES AND WE ARE GLAD TO WELCOME HIS GOSPEL ]|
* +------------------------------------------------------------------------------------------+
* }}}
*
* @param uid
* required uid for storing annotator to disk
* @groupname anno Annotator types
* @groupdesc anno
* Required input and expected output annotator types
* @groupname Ungrouped Members
* @groupname param Parameters
* @groupname setParam Parameter setters
* @groupname getParam Parameter getters
* @groupname Ungrouped Members
* @groupprio param 1
* @groupprio anno 2
* @groupprio Ungrouped 3
* @groupprio setParam 4
* @groupprio getParam 5
* @groupdesc param
* A list of (hyper-)parameter keys this annotator can take. Users can set and get the
* parameter values through setters and getters, respectively.
*/
class HubertForCTC(override val uid: String) extends Wav2Vec2ForCTC(uid) {
/** Annotator reference id. Used to identify elements in metadata or to refer to this annotator
* type
*/
def this() = this(Identifiable.randomUID("HubertForCTC"))
override def onWrite(path: String, spark: SparkSession): Unit = {
super.onWrite(path, spark)
getEngine match {
case TensorFlow.name =>
writeTensorflowModelV2(
path,
spark,
getModelIfNotSet.tensorflowWrapper.get,
"_hubert_ctc",
HubertForCTC.tfFile,
configProtoBytes = getConfigProtoBytes)
case ONNX.name =>
writeOnnxModel(
path,
spark,
getModelIfNotSet.onnxWrapper.get,
"_hubert_ctc",
HubertForCTC.onnxFile)
}
}
}
trait ReadablePretrainedHubertForAudioModel
extends ParamsAndFeaturesReadable[HubertForCTC]
with HasPretrained[HubertForCTC] {
override val defaultModelName: Some[String] = Some("asr_hubert_large_ls960")
/** Java compliant-overrides */
override def pretrained(): HubertForCTC = super.pretrained()
override def pretrained(name: String): HubertForCTC = super.pretrained(name)
override def pretrained(name: String, lang: String): HubertForCTC =
super.pretrained(name, lang)
override def pretrained(name: String, lang: String, remoteLoc: String): HubertForCTC =
super.pretrained(name, lang, remoteLoc)
}
trait ReadHubertForAudioDLModel extends ReadTensorflowModel with ReadOnnxModel {
this: ParamsAndFeaturesReadable[HubertForCTC] =>
override val tfFile: String = "hubert_ctc_tensorflow"
override val onnxFile: String = "hubert_ctc_onnx"
def readTensorflow(instance: HubertForCTC, path: String, spark: SparkSession): Unit = {
instance.getEngine match {
case TensorFlow.name =>
val tf = readTensorflowModel(path, spark, "_hubert_ctc_tf", initAllTables = false)
instance.setModelIfNotSet(spark, Some(tf), None)
case ONNX.name =>
val onnxWrapper =
readOnnxModel(path, spark, "_hubert_ctc_onnx", zipped = true, useBundle = false, None)
instance.setModelIfNotSet(spark, None, Some(onnxWrapper))
}
}
addReader(readTensorflow)
def loadSavedModel(modelPath: String, spark: SparkSession): HubertForCTC = {
val (localModelPath, detectedEngine) = modelSanityCheck(modelPath)
val vocabJsonContent = loadJsonStringAsset(localModelPath, "vocab.json")
val vocabJsonMap =
parse(vocabJsonContent, useBigIntForLong = true).values
.asInstanceOf[Map[String, BigInt]]
val preprocessorConfigJsonContent =
loadJsonStringAsset(localModelPath, "preprocessor_config.json")
val preprocessorConfig =
Preprocessor.loadPreprocessorConfig(preprocessorConfigJsonContent)
/*Universal parameters for all engines*/
val annotatorModel = new HubertForCTC()
.setVocabulary(vocabJsonMap)
.setDoNormalize(preprocessorConfig.do_normalize)
.setFeatureSize(preprocessorConfig.feature_size)
.setPaddingSide(preprocessorConfig.padding_side)
.setPaddingValue(preprocessorConfig.padding_value)
.setReturnAttentionMask(preprocessorConfig.return_attention_mask)
.setSamplingRate(preprocessorConfig.sampling_rate)
annotatorModel.set(annotatorModel.engine, detectedEngine)
detectedEngine match {
case TensorFlow.name =>
val (wrapper, signatures) =
TensorflowWrapper.read(localModelPath, zipped = false, useBundle = true)
val _signatures = signatures match {
case Some(s) => s
case None => throw new Exception("Cannot load signature definitions from model!")
}
/** the order of setSignatures is important if we use getSignatures inside
* setModelIfNotSet
*/
annotatorModel
.setSignatures(_signatures)
.setModelIfNotSet(spark, Some(wrapper), None)
case ONNX.name =>
val onnxWrapper =
OnnxWrapper.read(spark, localModelPath, zipped = false, useBundle = true)
annotatorModel
.setModelIfNotSet(spark, None, Some(onnxWrapper))
case _ =>
throw new Exception(notSupportedEngineError)
}
annotatorModel
}
}
/** This is the companion object of [[HubertForCTC]]. Please refer to that class for the
* documentation.
*/
object HubertForCTC extends ReadablePretrainedHubertForAudioModel with ReadHubertForAudioDLModel