Question / Help Horrendous Audio Quality and no idea why...

[Kaiman]

I have everything set to good levels and have even tried turning down the video quality to see if it would help, but nothing I have tinkered with seems to change the audio sounding like it is coming out of an earbud you have taken out of your ear. I'm running Windows 8 and have no problem with my stream at high quality other than the audio being awful. I submitted my log to the auto thing and it had no problems with anything audio-related.

 

[Blackhive]

post log here.
also check if your audio is set to aac 160k or mp3 128 ish(can go higher). aac 160k is preferable i think

 

[Kaiman]

Ok, I figured out what is wrong. The audio driver or whatever works terribly or something with OBS. When I used my headphones and set the stream to use that audio as the output it sounded perfect. Really weird that, but the mic built in with my computer had a decent output audio as well even though I am pretty sure it uses the same software as my audio.

Here are my logs, though, because I would love to not have to use my headphones to get decent audio.

20:46:23: Open Broadcaster Software v0.592b - 32bit (´・ω・｀)
20:46:23: -------------------------------
20:46:23: CPU Name: Intel(R) Core(TM) i7-4700HQ CPU @ 2.40GHz
20:46:23: CPU Speed: 2394MHz
20:46:23: Physical Memory:  4095MB Total, 4095MB Free
20:46:23: stepping id: 3, model 60, family 6, type 0, extmodel 1, extfamily 0, HTT 1, logical cores 8, total cores 4
20:46:23: monitor 1: pos={0, 0}, size={1920, 1080}
20:46:23: Windows Version: 6.2 Build 9200 
20:46:23: Aero is Enabled
20:46:23: -------------------------------
20:46:23: OBS Modules:
20:46:23: Base Address     Module
20:46:23: 001C0000         OBS.exe
20:46:23: 69A50000         OBSApi.dll
20:46:23: 699E0000         DShowPlugin.dll
20:46:23: 699C0000         GraphicsCapture.dll
20:46:23: 699A0000         NoiseGate.dll
20:46:23: 69980000         PSVPlugin.dll
20:46:23: ------------------------------------------
20:46:23: Adapter 1
20:46:23:   Video Adapter: NVIDIA GeForce GTX 780M
20:46:23:   Video Adapter Dedicated Video Memory: 3221225472
20:46:23:   Video Adapter Shared System Memory: 1073676288
20:46:23:   Video Adapter Output 1: pos={0, 0}, size={1920, 1080}, attached=true
20:46:23: =====Stream Start: 2014-01-14, 20:46:23===============================================
20:46:23:   Multithreaded optimizations: On
20:46:23:   Base resolution: 1920x1080
20:46:23:   Output resolution: 1920x1080
20:46:23: ------------------------------------------
20:46:23: Loading up D3D10 on NVIDIA GeForce GTX 780M (Adapter 1)...
20:46:23: ------------------------------------------
20:46:23: Audio Format: 48000hz
20:46:23: Playback device Default
20:46:23: ------------------------------------------
20:46:23: Using desktop audio input: Speakers (Realtek High Definition Audio)
20:46:23: ------------------------------------------
20:46:23: Using auxilary audio input: Microphone (Realtek High Definition Audio)
20:46:23: ------------------------------------------
20:46:23: Audio Encoding: AAC
20:46:23:     bitrate: 128
20:46:23: Using graphics capture
20:46:23: Scene buffering time set to 700
20:46:23: ------------------------------------------
20:46:23: Video Encoding: x264
20:46:23:     fps: 60
20:46:23:     width: 1920, height: 1080
20:46:23:     preset: veryfast
20:46:23:     profile: high
20:46:23:     keyint: 250
20:46:23:     CBR: yes
20:46:23:     CFR: yes
20:46:23:     max bitrate: 1000
20:46:23:     buffer size: 1000
20:46:23: ------------------------------------------
20:46:23: MMDeviceAudioSource: Frequency for device 'Speakers (Realtek High Definition Audio)' is 1536000, samples per sec is 192000
20:46:23: MMDeviceAudioSource: Frequency for device 'Microphone (Realtek High Definition Audio)' is 352800, samples per sec is 44100
20:46:25: Using RTMP service: Twitch / Justin.tv
20:46:25:   Server selection: rtmp://live-iad.justin.tv/app
20:46:25:   Interface: Killer Wireless-N 1202 (2.4GHz and 5GHz) (802.11, 144 mbps)
20:46:26: Completed handshake with rtmp://live-iad.justin.tv/app in 211 ms.
20:46:26: SO_SNDBUF was at 65536
20:46:26: SO_SNDBUF is now 65536
21:04:57: RTMPPublisher::SocketLoop: Stalled for 1846 ms to write 143337 bytes (buffer: 0 / 144384), unstable connection?
21:05:22: Total frames encoded: 68284, total frames duplicated: 184 (0.27%)
21:05:22: Total frames rendered: 68211, number of late frames: 39 (0.06%) (it's okay for some frames to be late)
21:05:22: RTMPPublisher::SocketLoop: Graceful loop exit
21:05:22: Average send payload: 2774 bytes, average send interval: 21 ms
21:05:22: Number of times waited to send: 2, Waited for a total of 4114 bytes
21:05:22: Number of b-frames dropped: 56 (0.082%), Number of p-frames dropped: 56 (0.082%), Total 112 (0.16%)
21:05:22: Number of bytes sent: 147818694
21:05:22: 
21:05:22: Profiler time results:
21:05:22: 
21:05:22: ==============================================================
21:05:22: video thread frame - [100%] [avg time: 0.655 ms] [children: 11.6%] [unaccounted: 88.4%]
21:05:22: | scene->Preprocess - [0.153%] [avg time: 0.001 ms]
21:05:22: | GPU download and conversion - [11.5%] [avg time: 0.075 ms] [children: 9.92%] [unaccounted: 1.53%]
21:05:22: | | flush - [5.65%] [avg time: 0.037 ms]
21:05:22: | | CopyResource - [3.82%] [avg time: 0.025 ms]
21:05:22: | | conversion to 4:2:0 - [0.458%] [avg time: 0.003 ms]
21:05:22: Convert444Threads - [100%] [avg time: 1.425 ms] [children: 99.4%] [unaccounted: 0.632%]
21:05:22: | Convert444toNV12 - [99.4%] [avg time: 1.416 ms]
21:05:22: encoder thread frame - [100%] [avg time: 1.734 ms]
21:05:22: ==============================================================
21:05:22: 
21:05:22: 
21:05:22: Profiler CPU results:
21:05:22: 
21:05:22: ==============================================================
21:05:22: video thread frame - [cpu time: avg 0.175 ms, total 12000 ms] [avg calls per frame: 1]
21:05:22: | scene->Preprocess - [cpu time: avg 0 ms, total 0 ms] [avg calls per frame: 1]
21:05:22: | GPU download and conversion - [cpu time: avg 0.043 ms, total 2968.75 ms] [avg calls per frame: 1]
21:05:22: | | flush - [cpu time: avg 0.023 ms, total 1578.13 ms] [avg calls per frame: 1]
21:05:22: | | CopyResource - [cpu time: avg 0.014 ms, total 984.375 ms] [avg calls per frame: 1]
21:05:22: | | conversion to 4:2:0 - [cpu time: avg 0.002 ms, total 187.5 ms] [avg calls per frame: 1]
21:05:22: Convert444Threads - [cpu time: avg 1.359 ms, total 185328 ms] [avg calls per frame: 2]
21:05:22: | Convert444toNV12 - [cpu time: avg 1.354 ms, total 184703 ms] [avg calls per frame: 2]
21:05:22: encoder thread frame - [cpu time: avg 1.073 ms, total 73234.4 ms] [avg calls per frame: 1]
21:05:22: ==============================================================
21:05:22: 
21:05:22: =====Stream End: 2014-01-14, 21:05:22=================================================

 

[Blackhive]

make sure you have the microphone on a push to talk hotkey or it could pickup whatever sound is coming from your speakers, which would then mean your audience would hear game music/sound+ game music/sound echoed through your microphone, in addition to you talking.

also, if you want to stream 1920x1080, you want 30fps. twitch's bitrate max is 3500 for partners, 2500 for non-partners. if your max bitrate is less than 5mbps forget about streaming at 1920x1080, and definately forget about 60fps. so.. some quick general rules regarding your system, not others:

twitch-partner:
+5mbps upload==
video: 1920x1080 30fps 3500 bitrate/buffer with cbr +cbr padding
or
video: Base resolution: 1920x1080>resolution downscale 1280x720 60fps 3500 bitrate/buffer with cbr +cbr padding
audio: aac 160k

non-partner:
video: Base resolution: 1920x1080>resolution downscale 1280x720 30fps 2500 bitrate/buffer with cbr +cbr padding
audio: aac 160k

if you have <4mbps upload speed you can also use 2000 or even 1500 bitrate.

 

[Sapiens]

MMDeviceAudioSource: Frequency for device 'Speakers (Realtek High Definition Audio)' is 1536000, samples per sec is 192000

Try toning that down to something like 16 bit 48 Hz. 24 bit 192 Hz is a bit overkill and may be contributing to the distortion problem.