Help needed recording Vinyl, results sound harsh


I’ve recently attempted to record my record collection with mixed results.

I’m using a very well set up TT into a standalone phono stage which in turn is then connected to a Cakewalk UA1G USB convertor, hooked up to my laptop using its own battery supply.

I then ripp to Audacity using 36bit 96khz, I then amplify, high pass filter before exporting as a WAV.

I then declick, import back to Audacity for noise removal and then export again as a 16bit 41khz WAV before reimporting to splice tracks and then finally export to Flac.

The trouble is no matter what I have tried the result always sounds hard compared to the vinyl when it is played conventionally on my system. All the sweetness is lost and the sound is fatiguing and difficult to listen to.

I’ve downloaded other peoples vinyl ripps that sound sweet on my system so I know it can be done. I’m just not sure what I need to do or change?

Would there be much difference in the AD/DA convertors, would I be better off trying a different one?

Any ideas would be much apreciated.

I “rip” vinyl with the predecessor device to the UA-1g that you have (The Ediril UA-1EX) it has always produced excellent results.

Why are you using such an insanely high Hz setting at 96 kHz ? I suspect that the “harshness” you are hearing may be down to the downsampling 96=>44.1kHz - but don’t quote me on that :wink:

I record in Audacity with Audacity set to 32-bit floating 44.1kHz (the 32-bit gives good headroom for editing) - at final stage of production I export to 16-bit PCM stereo 44.1kHz WAV for loading into iTunes and/or making a CD. I have my dither (applied on downsampling to “smooth” the signal) set to “Triangular”.

I used to run my UA-1EX in its “Standard” mode with the default windows drivers - lately I have it set in “Advanced” mode using the Edirol drivers downloaded from their website. It has produced excellent results in both modes.

When recording I aim for a maximum peak signal of around -6.0dB (50% if you are in linear mode on the waveform or meters). Then after all the editing I use Effect>Amplify to lift the signal level to a max. -2.0dB (some players/software don’t play well with signals that peak at 0dB - and -2dB is plenty loud enough).

Tip: you can make you metering more accurate bt enlarging the meters, click and drag on the meter toolbar to resize it.


The harshness (too much treble?) could be due to lack of RIAA de-emphasis. Audacity has a preset curve in the equalizer to do this …

… most records manufactured from the 1950s onwards are produced with RIAA equalization, a form of pre-emphasis which boosts high frequencies and reduces low frequencies, which is then de-emphasized on playback … it’s possible to directly connect a standalone turntable to the line input of a high quality sound card, if you are prepared to perform the amplification and RIAA equalization yourself in Audacity. Effect > Equalization has a suitable RIAA preset.

RIAA de-emphasis curve in Audacity equalizer, (cut treble, boosts bass).png

Hey WC,

Thanks so much for your advice.

I’ve followed your settings (I also resett the jumpers on the back of the Cakewalk) and it sounds much improved, hardness has gone and harmonics are improved as well :smiley:

With respect to the Quality settings in Audacity should I set both the ‘Real time conversion’ and the ‘High quality conversion’ settings to High quality synch and dither to triangle?

Thanks again


these are my Quality settings:
my Quality preferences.PNG
With regards to the Dither settings, have a read of this article in the Wiki:


“High Quality Sync Interpolation” and “Triangle dither” (or “shaped dither”) are good for the “High Quality Conversion” settings, but if used for the “Real Time Conversion” settings will tend to make the computer sluggish and reduce Audacity’s performance. The “Fast” and “No dither” settings (as per waxcylinders screenshot) are perfectly adequate for Real Time Conversion and have no effect on the actual recording or the finished product.

Thanks all,

I have made some further improvements to the sound quality by reducing the Impedance on my moving coil phono stage.

I’m currently listening to some test WAVs recorded at 24bit/48hz and they sound very good.

However, when I downsample them to Flac they lose a bit of the sparkle.

I’ve also noticed that any kind of post processing seems to degrade the sound to a degree as well.

I may just stick with the 24/48 WAVs as they sound best and I have a 1TB drive on my Squeezebox so I could still get a fair few LP’s on there!

You may be unintentionally saving as 16 bit when you save as FLAC …
I think the default setting for FLAC in Audaciity  is 16 bit (rather than 24 bit).png

48 kHz is the standard for DVD’s. CD’s standard is 44.1 kHz. If you plan on writing it to audio CDs later you should use 44.1 kHz instead of 48kHz. Resampling from 48 kHz to 44.1 kHz (or the other way) is not recommended. Those 2 frequencies were choosen on purpose to dificult conversion from one to the other.

Have you ever tried an ABX test on that (double blind comparison)? With a good resampling algorithm the conversion is extremely close. With the highest quality conversion using SoX the conversion is near perfect.

Actually no, but I probably wouldn’t tell the difference, I don’t have audiophile ears! I’ve tried blind tests on lossless vs lossy formats and I couldn’t really tell the difference where other have, so I guess I might not be the best subject for that kind of test :slight_smile:

In theory (mathematically speaking) conversion between those 2 sample rates is supposed to be “not-the-best”. If not for any other reason, avoiding one extra step on the process might save some time.

oK, settled on 24/44.1KHZ WAVs

I seem to consistently prefer these to 16/44.1khz wavs on doubleblind tests.

Q, are there better wav to Flac convertors than those in Audacity?

I saw Sox mentioned.

Better in which way? Faster? Capable of batch processing? You’ll probably find plenty of them, some being free software. SoX is a command line tool, if the goal is to simply convert from wav to flac there might be better (and by better here I mean “prettier”) options. I’m not a windows user so I can’t really recommend any wav-flac converter for windows, but you can find some links in the FLAC website: (scroll down to the “extras” section).

From my experience there is no difference in speed or sound quality whichever Flac converter you use. The only real difference is the graphical interface, underneath which they all use the same encoder (though some may use older or newer versions by a point or two).

Just a quick update,

I reset the stream format decoders in my Squeezebox Duet so that the SB wasn’t doing the encoding locally and my god what a difference that has made to the SQ.

I can now play Flacs and not be able to distguish them from the WAVs.

I also lowered the input on my Cakewalk when recording from Vinyl so that audacity peaks at -6db and topped up the bearing oil on my TT.

The results I am getting now are excellent and I am finally really happy with the results.

I’ve noticed now that if I do some gentle post processing such as declick and noise reduction that the results do not detract from the music.

Very happy now and would recomend that anyone who uses a SB Duet resets their file format preferences and also anyone who uses a Cakewalk is very conservative with their input levels.