top of page
  • Writer's pictureSound Reformer / yamakaWA

Voice extraction of voice memos is difficult with compressed sound sources (mp3) with poor data de

Voice extraction of voice memos is difficult with compressed sound sources (mp3) with poor data density


I'm Yamakawa, a sound reformer who may be dealing with poor quality voice data in Japan. Since the sound quality improvement site I run has a free trial system, more than 100 low-quality sound sources that say "that kind of sound is impossible !!" are collected annually.

Among them, there are many worries about wanting to extract the voice of voice memos recorded with an IC recorder or smartphone, but to be clear, the probability of being adopted from a trial is about half. The probability of improvement without problems is quite low.

What kind of recording is difficult to improve?

  • In the first place, it is far from the sound source (I can only hear the sound of mosquitoes)

  • Conversations other than the people themselves are involved in izakaya and noisy cafes.

  • Make the faint voice of the other party that you can hear from the handset on the telephone.

The microphone and the amount of data determine the recording quality

Audio recording and video audio are digital, so they are digitized and recorded. That's why the microphone and recording resolution are important. Amateur recordings do not adhere to the basic principles. Only people who are recording without checking the recorded sound. Few people use earphones or headphones to check the sound on a monitor to see what kind of picture they are taking with a digital camera.

The only way to extract voice from noisy conversation data is to amplify or resynthesize the remaining components of the digitized data. If you want to make a good sound later, the resolution of audio material with a large amount of data is the key.

Imagine a digital camera image with a rough image. It's easy to imagine that a photo that looks blurry and looks like a mosaic is unlikely to be fixed. Voice data is basically the same.

IC recorder long-time recording (MP3)

MP3 files can be recorded for a long time because the amount of data can be saved. Many IC recorders can switch the recording method. (Wav and mp3) mp3 has a capacity of about 1/5 that of wav, so it is convenient for long-time recording. However, a large amount of data is being decimated.

It's okay to compress professional quality sound sources such as artist music data and listen to them in MP3, but it's not suitable for amateurs who want to record with high sound quality. In noisy environments, such as where the recorder's microphone is far from conversation or noisy, it is difficult to technically repair it later.

Hidden recording to suppress cheating scene

In many cases, one of the spouses reminds us of an IC recorder (smartphone) in a room or the like in an attempt to suppress evidence of an affair. In fact, this usage is wrong. Although it differs depending on the model, the IC recorders that are sold at low prices have a proper recording range of about 1 meter over the desk, such as for interviews. Even if it is hidden in the corner of the room, it cannot record decent audio.

If you need to record from a few meters away and listen to the details, you need a higher-grade recorder equipped with a high-performance microphone called a linear PCM recorder.


・ Check with headphones what kind of sound can be recorded before recording if it is an important voice that you want to keep as evidence.

・ If you want to leave the possibility of repairing later, record with wav method.

・ If you want to record a conversation with a voice memo on an IC recorder or smartphone, it is within 1 meter from the microphone.


bottom of page