search

n7tae / mvoice

A prototype M17 voice application for ham radio
GNU General Public License v2.0
55 stars 13 forks source link

low audio quality? #22

Closed sysgdc closed 2 months ago

sysgdc commented 2 months ago

I just built the latest mvoice ver 1.1.4 on Ubuntu 22.x = Linux Mint 21.1. No problems. Builds clean. It works.

Note my audio playback using mvoice Echo Test is intelligible but pretty low quality compared to other voice software. I am using a quality Plantronics USB headset. I am not sure if this is a limitation of Codec2? I've built mvoice with USE44100 = true but that doesn't seem to make any difference. (Echo Time=8.6s Vol=-3dB Clip=0%)

I don't consider the audio quality to be a big problem. I just don't have enough experience with m17 to know if there are parameters I can control to improve it or if that's 'normal'

73, Greg K1UH

n7tae commented 2 months ago

Codec2 should sound every bit as good as DStar or DMR, maybe even a bit better.

The README should say that you want Clip=0% and a Vol=-14dB -ish. Then adjust your receive volume for a comfortable level.

I use an inexpensive audio plug over-the-ear JBL headset. It has no mic gain adjustment. I set the mic gain in the Ubuntu Sound config. Once I get the clipping and vol right, I turn the headset ear-cup volume all the way up and then adjust my listening volume with my Ubuntu sound configuration.

Let me know if that sounds better and I'll change the README.

n7tae commented 2 months ago

I doubt if you need USE44100. That's for really old systems.

sysgdc commented 2 months ago

I cranked down my Ubuntu mixer microphone volume control to 58% (-14db) and the quality seems to be slightly improved. Echo Time=18.3s Vol=-16dB Clip=0% Echo Time=13.7s Vol=-12dB Clip=0%

Echo Test playback audio seems nearly equal to ~DMR quality but voice still sounds a bit synthetic to my ears, especially compared to wideband codecs like G.7xx, VOIP, etc.

Conclusion: I think the latest Operating description in the README.md is accurate. Okay with me to close this issue.

PS:

Note there is a minor typo link error in README, --> Building and installing the OpenDHT support https://github./com/savoirfairelinux/opendht.git

Regards,

73, Greg K1UH

On Tue, Aug 20, 2024 at 5:28 PM Tom Early @.***> wrote:

I doubt if you need USE44100. That's for really old systems.

— Reply to this email directly, view it on GitHub https://github.com/n7tae/mvoice/issues/22#issuecomment-2299796325, or unsubscribe https://github.com/notifications/unsubscribe-auth/ASZZUC7UWUNUCJDTUYOACVTZSOYGPAVCNFSM6AAAAABM2VFNXWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDEOJZG44TMMZSGU . You are receiving this because you authored the thread.Message ID: @.***>

n7tae commented 2 months ago

Thanks for the typo detection.

M17 voice-only is 3200 bps and voice+data is 1600 bps. Either way, M17 is not a wide band method. To my ears voice-only is better than DStar or DMR/YSF. Voice+data is not as good but still works quite well.