Initial commit: FunASR Speech Recognition Toolkit
Update API Documentation / build-api-docs (push) Has been cancelled
Update API Documentation / build-api-docs (push) Has been cancelled
Add complete FunASR codebase including models, runtime, and documentation.
This commit is contained in:
@@ -0,0 +1,108 @@
|
||||
// itf/transition-information.h
|
||||
|
||||
// Copyright 2021 NVIDIA (author: Daniel Galvez)
|
||||
|
||||
// See ../../COPYING for clarification regarding multiple authors
|
||||
//
|
||||
// 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
|
||||
//
|
||||
// THIS CODE IS PROVIDED *AS IS* BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
|
||||
// KIND, EITHER EXPRESS OR IMPLIED, INCLUDING WITHOUT LIMITATION ANY IMPLIED
|
||||
// WARRANTIES OR CONDITIONS OF TITLE, FITNESS FOR A PARTICULAR PURPOSE,
|
||||
// MERCHANTABLITY OR NON-INFRINGEMENT.
|
||||
// See the Apache 2 License for the specific language governing permissions and
|
||||
// limitations under the License.
|
||||
|
||||
#ifndef KALDI_ITF_TRANSITION_INFORMATION_H_
|
||||
#define KALDI_ITF_TRANSITION_INFORMATION_H_
|
||||
|
||||
#include <stdint.h>
|
||||
#include <vector>
|
||||
|
||||
namespace kaldi {
|
||||
|
||||
/**
|
||||
* Class that abstracts out TransitionModel's methods originally used
|
||||
* in the lat/ directory. By instantiating a subclass of this abstract
|
||||
* class other than TransitionModel, you can use kaldi's lattice tools
|
||||
* without a dependency on the hmm/ and tree/ directories. For
|
||||
* example, you can consider creating a subclass that implements these
|
||||
* via extracting information from Eesen's CTC T.fst object.
|
||||
*
|
||||
* TransitionId values must be contiguous, and starting from 1, rather
|
||||
* than 0, since 0 corresponds to epsilon in OpenFST.
|
||||
*/
|
||||
class TransitionInformation {
|
||||
public:
|
||||
virtual ~TransitionInformation() {};
|
||||
/**
|
||||
* Returns true if trans_id1 and trans_id2 can correspond to the
|
||||
* same phone when trans_id1 immediately precedes trans_id2 (i.e.,
|
||||
* trans_id1 occurss at timestep t, and trans_id2 ocurs at timestep
|
||||
* 2) (possibly with epsilons between trans_id1 and trans_id2) OR
|
||||
* trans_id1 ocurs before trans_id2, with some number of
|
||||
* trans_id_{k} values, all of which fulfill
|
||||
* TransitionIdsEquivalent(trans_id1, trans_id_{k})
|
||||
*
|
||||
* If trans_id1 == trans_id2, it must be the case that
|
||||
* TransitionIdsEquivalent(trans_id1, trans_id2) == true
|
||||
*/
|
||||
virtual bool TransitionIdsEquivalent(int32_t trans_id1, int32_t trans_id2) const = 0;
|
||||
/**
|
||||
* Returns true if this trans_id corresponds to the start of a
|
||||
* phone.
|
||||
*/
|
||||
virtual bool TransitionIdIsStartOfPhone(int32_t trans_id) const = 0;
|
||||
/**
|
||||
* Phone is a historical term, and should be understood in a wider
|
||||
* sense that also includes graphemes, word pieces, etc.: any
|
||||
* minimal entity in your problem domain which is represented by a
|
||||
* sequence of transitions with a PDF assigned to each of them by
|
||||
* the model. In this sense, Token is a better word. Since
|
||||
* TransitionInformation was added to subsume TransitionModel, we
|
||||
* did not want to change the call site of every
|
||||
* TransitionModel::TransitionIdToPhone to
|
||||
* TransitionInformation::TransitionIdToToken.
|
||||
*/
|
||||
virtual int32_t TransitionIdToPhone(int32_t trans_id) const = 0;
|
||||
/**
|
||||
* Returns true if the destination of any edge with this trans_id
|
||||
* as its ilabel is a final state (or if a final state is
|
||||
* epsilon-reachable from its destination state).
|
||||
*/
|
||||
virtual bool IsFinal(int32_t trans_id) const = 0;
|
||||
/**
|
||||
* Returns true if *all* of the FST edge labeled by this trans_id
|
||||
* have the same start and end states.
|
||||
*/
|
||||
virtual bool IsSelfLoop(int32_t trans_id) const = 0;
|
||||
int32_t TransitionIdToPdf(int32_t trans_id) const {
|
||||
return TransitionIdToPdfArray()[trans_id];
|
||||
}
|
||||
/**
|
||||
* Returns the contiguous array that backs calls to
|
||||
* TransitionIdToPdf().
|
||||
*
|
||||
* Ideally, this would return a std::span, but it doesn't because
|
||||
* kaldi doesn't support C++20 at the time this interface was
|
||||
* written.
|
||||
*/
|
||||
virtual const std::vector<int32_t>& TransitionIdToPdfArray() const = 0;
|
||||
int32_t NumTransitionIds() const {
|
||||
return TransitionIdToPdfArray().size() - 1;
|
||||
}
|
||||
/**
|
||||
* Return the number of distinct outputs from
|
||||
* TransitionIdToPdf(). Another way to look at this is as the number
|
||||
* of outputs over which your acoustic model does a softmax.
|
||||
*/
|
||||
virtual int32_t NumPdfs() const = 0;
|
||||
};
|
||||
|
||||
} // namespace kaldi
|
||||
|
||||
#endif // KALDI_TRANSITION_INFORMATION_H_
|
||||
Reference in New Issue
Block a user