Voice clarification guidance request

A family member was hospitalized and succumbed to her injuries two days later. An investigation was opened and closed the same day. When her family saw that there would be no investigation we began to gather as much information as possible with the goal of having authorities reopen the investigation (best outcome) or pursuing a civil suit. We found in her medical records a documented history of ongoing abuse which she had hidden from her siblings and other family members. When the family began asking questions, friends and coworkers came forward to tell us that she had confided in them that she was suffering ongoing abuse.

The victim left a voicemail 1 hour and 5 minutes before an ambulance was dispatched to bring her to the hospital. The voicemail is the last known communication from her. It lasts exactly 4 minutes, which I assume is the system limit set by the cell phone provider. No one seems to speak into the mic of the phone and the voices are distorted. As a life-long IT professional who dabbles in live sound reinforcement, I was sent a copy to see if I could glean any information from the recording. To me, the voices sound like an overdriven mic, but the opposite makes more sense. Fortunately, there is nothing horrifying recorded as I feared the first time listening through.

Using Audacity, I’ve spent hours using different effects…compressor, EQ, noise gate, vocal isolation, and other effects to make the voices understandable. I’ve had some success and am able to hear some conversation. If I’m hearing correctly, it is significant. I’ve reached out to forensics labs and their minimum prices are out of reach for us currently.

My questions: First, am I wasting my time attempting to make a recording from a (possibly) underdriven microphone understandable? Second, if it can be done, can someone point me towards the right effects/tools/process to clean up the voice? Third, is anyone interested in taking this on as a project, either paid or as a kindness to a family seeking justice?

We aren’t worried about forensic documentation or court testimony. We only want to hear what is said…to know if she left us an important message or if it was a butt dial.

The original file is an .amr file pulled from the recipient’s phone. I’ve converted that to a .wav file using VLC media player. I shared these in a Google Drive folder in case taking a listen will help. The original and the converted .wav file are shared here: Original Voicemail - Google Drive

Thanks for reading and any help is greatly appreciated.

Can try this for free …

NB: AI is capable of confabulation.

Don’t get your hopes up: if you cannot understand what is being said,
processing is probably not going to make it comprehensible, (that only happens in the movies).

Thank you, Trebor. I had given the podcast tool a shot using both the original and my best processed version. It seems like I got some improvement with the Enhance strength set between 2% and 4%, but with the strength set any higher than that it was all “garbage out” as you predicted. I will likely try again if I can get a copy any cleaner than what I’ve come up with.

I also used the Auphonic app online which has some really nice automated features.

This topic was automatically closed after 30 days. New replies are no longer allowed.