Hi, I have a client who has some pretty old audio recording devices that output audio in a custom format so you have to use their software, but classic case, the company that sells the devices has gone under many years ago and there is no documentation.
I have been able to extract the raw audio from their format but have not been able to decode it trying every combination under the sun and was hoping I could get some pointers or help here.
The file I have is of someone counting from 1 to 10 over a duration of approximately 9 seconds (I think it is about 9.8ish). According to the recording, it was done using 16bit at 22050 hz as stereo/2 channel.
I have tried decoding it using those settings as PCM and I get a 9 second audio file but all I hear is static. I then tried VOX ADPCM at 192000 hz and it seem to come out with the static again but I think I can hear a faint counting in the background. When I changed it to mu/ulaw as 1 channel with a range of 96000hz the counting seems slightly more apparent but there is till lots of static.
On the device’s spec sheet it says Quantisation is 16 bit LPCM Linear / 8 bit ADPCM Non-Linear 4 bit ADPCM Non-Linear but when I try and use FFMPEG to decode it using an ADPCM format I get an error and it won’t decode.
So I was hoping someone may be able to point me in the right direction or even explain why the different sampling rates for VOX and ulaw provide the correct duration of the file and somewhat better quality.
It would be useful to know what the device in question is and what the software they provided is called. Are you able to provide that information, please?
Unfortunately I don’t have much info, it was called an Alpha-A2 and is European. The client does not have anyone around who knows the background anymore but have a large archive of files they wanted to convert.
If it is an audio device, it will lsurely have the capability to output the sound to an amplifier or to a loudspeaker. At best, it has a line-out connector, and you can use a cable and an external sound card connected to the USB port of a computer. Or, if the computer has a line-in, you don’t even need an external sound card. Please note that a microphone input on the computer is not the same as a line-in.
I managed to find this site with several audio devices called Alpha-Something: Juno. I don’t know if it’s the same company but they might be a place to start. The other Alpha A2 things I Googled were a pressure washer and some medical condition.
Decoding RAW audio from an old device can be challenging, but it’s definitely manageable. First, identify the audio format (e.g., PCM, ADPCM) used by the device. You may need specialized software like Audacity or VLC Media Player that supports various audio codecs. Once you have the right tools, you can import the RAW audio data and specify parameters like sample rate and bit depth for successful playback. If you encounter specific issues, let me know, and I can help troubleshoot!
Thanks for all the replied will try and respond to provide more info.
It is a specialised device that records data to an internal SD card that was then uploaded to specialised software to get the data off. It was used in a legal firm to record conversations in meeting rooms and has no audio outputs or USB connectivity. It is designed to log audio and download the data to be catalogued.
I have access to the the files it creates, from which I have extracted the raw audio at posted above.
We have done lots of searching, the company no longer exists (Audnautics) and we cannot find anything online - we spent many hours trying to find something.
As specified above I know the recording rates and sampling, and the device has a spec sheet stating 6 bit LPCM Linear / 8 bit ADPCM Non-Linear 4 bit ADPCM Non-Linear, so I assume it is ADPCM, however I have been unable to convert it using Audacity, Sox or FFMPEG. I have also tried decoding it as PCM/ulaw/alway/vox adpcm and they all come out with lots of static (although ulaw/vox adpcm does have a very faint sound of someone counting within the static - as you can listen to in the link I posted above).
Importing it as 16kbps NMS ADPCM (LE) as 1 channel, 22050Hz sample does render some kind of audio but I would guess the device used some proprietary tweaks in addition to ADPCM encoding. FFMpeg should be able to render out what ever is fed to it (without crashing) but trying to guess the format would probably require writing a batch script that goes through the possible combinations, as there are a few.
Just to close this off, it seems to be a proprietary format so I don’t think any of the main tools are going to help.
We finally got the go ahead from our client to reverse engineer the legacy app they used to read the files so we are going to go in that direction now. Thanks for all the advice.
