International Audio Laboratories Erlangen

Lecture: Speech Enhancement, Winter Term 2020/2021

  • Instructor: Prof. Dr. Emanuël Habets
  • Teaching Assistant: Wolfgang Mack
  • Time: Winter Term 2020/2021, Thursday 08:15-09:45
  • Place: Zoom Meeting
  • Format: Lecture
  • Credits: 2,5 ECTS
  • Exam (graded): Oral examination at the end of the term
  • Flyer: PDF

Format

  • Due to the COVID-19 pandemic, the lecture will be offered as an online course via ZOOM.
  • The next online lecture will be on Thursday, January 14th, at 08:15.
  • Participation in the ZOOM session is only possible for FAU students. The ZOOM access information for this course will be made available via StudOn. Therefore, you must register via StudOn prior to the first lecture.
  • To ensure privacy, participants are not permitted to record the ZOOM sessions. Furthermore, ZOOM links may not be distributed.
  • Rather than following the traditional lecturing format, this course will be inspired by the flipped classroom concept. Being offered in this format for the first time, the lecture will have some experimental character.
    Important elements are:
    • Reading assignments.
    • Watching online videos.
    • Programming using Python.
    In particular, students are required to be prepared prior to the lecture. The lecture time will be used to provide a short summary, to deepen the most important aspects, and for having a question–answering dialogue with participants. Note that this concept will require a lot of work and dedication on the side of the lecturer and participants.
  • As a technical requirement, all participants must have access to a computer capable of running the ZOOM video conferencing software (as provided by FAU), including audio and video transmission as well as screensharing. Furthermore, a regular web browser (preferably Google Chrome) to access the Python development environment is needed.
  • The required material will be made available to the course participants in the following way:
    • All FAU students can get an electronic copy of the required reading material.
    • The slides used in the lecture are made available as PDF.
    • The Jupyter notebooks and audio examples are made available.
    • All required videos are made available.
    • Questions and answers during the ZOOM sessions will be collected and made accessible to course participants.

For further information, please contact Prof. Dr. Emanuël Habets.

Content

We live in a noisy world! In all applications that are related to speech from hands-free communication, teleconferencing, hearing aids, cochlear implants, to human-machine interfaces such as smart speakers, a speech signal of interest captured by one or more microphones is contaminated by noise and reverberation. Depending on the level of noise and reverberation, the quality and intelligibility of the captured speech can be greatly reduced. Therefore, it is highly desirable, and sometimes even indispensable, to "clean up" the noisy signals using signal processing techniques before storage, transmission or reproduction.

In this course both traditional and deep learning methods for noise reduction and dereverberation, with one or multiple microphones, are discussed.

The goal of this course is to provide a strong foundation for researchers, engineers, and graduate students who are interested in the problem of signal and speech enhancement.

Course Material

The lecture slides and videos can be downloaded on StudOn.

Exercises

Jupyter notebooks have been created that go with the exercises. To access them you need to

  1. Connect to the FAU network (directly or via VPN).
  2. Access juplab.audiolabs.uni-erlangen.de.
  3. Login with the provided user account.
  4. Select Python in “Spawner Options”. Make sure Launch “JupyterLab (beta)” is NOT checked and then click "Spawn".
  5. Start the Jupyter notebook with the file extension ipynb.

Links

Further audio-related courses offered by the AudioLabs can be found at: