Benchmark the CTC against a RNN-Transducer model

Description

This issue does not include a description.

Contributor guide

Tech stack

cpppythontensorflow

Domain

machine learning

Issue type

research

DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.

4

Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.

3-5 days

Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.

stale

ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.

unclear

Prerequisites

Understanding of CTC lossFamiliarity with RNN TransducerBasic knowledge of DeepSpeech codebase

Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.

15

Research direction

This issue asks to benchmark the Connectionist Temporal Classification (CTC) model against a RNN Transducer model within the DeepSpeech framework. Since the issue lacks description, one must first understand the current implementation (see DeepSpeech's model definition files, likely in Python). Then, implement a RNN Transducer model (e.g., using TensorFlow) and compare training time, word error rate, and inference speed. Existing discussions or linked PRs may provide context; check comments #753 for any hints. The benchmark should be reproducible and documented.