Sonntag, 19. Juli 2020

LIve Online Stream Mixdown Session

It wasn't just through Covid 19 that I came up with the idea of ​​offering my customers the opportunity to experience the mixdown of their title live over the Internet. Via YouTube he gets a look at my screen, in the studio (my person) and of course there is the live sound in a very good quality. Via chat there is the possibility to intervene live in the event and to help shape the sound. The customer hears his title in his familiar environment and can thus judge how the title sounds to his taste. The prerequisite for this is an Internet-capable computer with an audio card, connected speakers (or headphones) and a YouTube registration for the chat. Here's how it works: An appointment is made in advance. The customer will receive a YouTube URL by email on this date. If he enters / copies this URL into his internet browser, he will hear his title directly and be able to see the mixdown. He can contact me via chat via the Youtube platform. In most cases I will not start the mixdown from scratch, but there will be a pre-mix that can be played. In the course of the title, he can now express change requests that I implement directly. The session ends when the customer has found his sound. Stefan Noltemeyer

Montag, 13. Juli 2020

Youtube and LUFS (loudness units relative to full scale)

top:original, bottom: youtube version

What does youtube do with Loudness War Tracks, what happens to undersized files? I did an experiment. I play the same music three times with different levels (one minute running time). Original level: 1st version (top left in the picture) -26.5 LUFS with -10.6dBFS (True Peak) 2nd version (above middle in the picture) -16.5 LUFS with - 0.7 dBFS (True Peak) 3rd version (top right in the picture) -9.1 LUFS with overload + 0.5dBFS (True Peak) What does YouTube do with it? 1st version (bottom left of the picture) -21.0 LUFS with -5.9 dB 2nd version (bottom center of the picture) -14.6 LUFS with -0.7dB 3rd version (bottom right in the picture) -14.2 LUFS with - 4dB Conclusion here: It is striking that versions 2 and 3 have almost identical loudness, although the difference was originally more than 6 dB. It is particularly striking that the quiet version was only slightly raised in level, but was not brought up to -14LUFS. Experiments with level tones or level tone series: In a Youtube video (original level with 11.1 LUFS): 1KHz -20dB 1KHZ-10dB 10KHz -20dB 10KHz-10dB That resulted in the reproduction on Youtube: with 14.3 LUFS 1KHz -23dB 1KHZ-13dB 10KHz -23dB 10KHz-13dB next attempt (original level): with -1,8LUFS 1KHz -20dB 1KHZ-10dB 10KHz -20dB 10KHz-10dB 1KHz -1dB !! This resulted in the playback on Youtube: with 14.1 LUFS 1KHz -30dB 1KHZ-20dB 10KHz -30dB 10KHz-20dB 1KHz-15dB Conclusion here: Egel what level tones, what level, in the end it becomes about -14 LUFS. So the question arises, why wasn't the quiet audio file also brought to -14LUFS? Most music download platforms such as Spotify, Apple Music etc. operate in a similar way. The desired / implemented loudness is about -14 LUFS for fully modulated titles For this reason, we are now creating two masters, one with a loudness of -16 LUFS to -14 LUFS and one with the usual CD loudness. Stefan Noltem

the different vocal recording

I was faced with the task of recording a classical opera voice. Since I had already realized some rock / pop jazz productions with this singer, I proceeded as usual. My recording booth was perfect for vocal recordings and recordings of individual instruments like acoustic guitars or horns. The very low-reflection acoustics are good for a sound-neutral recording. 
Especially with quiet sound sources such as a spoken voice, otherwise a coloration is added to the original sound, which is ugly and irreversible. 
There is nothing to do against reflections from the walls, floor and ceiling. The audible coloring can not be clarified with any equalizer. Using a compressor in the mixdown,
reinforces the problem. Especially with quiet sound sources such as a spoken words coloration is added to the original sound, which is ugly and irreversible. 

As usual I used my setup with a Neumann TLM 49 microphone and a Focusrite ISA One preamp, but the high strong soprano voice didn´t sound that way we´re looking for. So we tried to raise up the distance to the mic up to one meter. The sound was a bit nicer, but has not really convinced. Our Internet research has shown that opera voices are often recorded with a stereo microphone at a distance of about 1.5 meters. Of course we tried that ... and we enjoyed the result.
The recording was placed in the 30 square meters control room. There is a reduced acoustics. with a RT60 of 0.3 sec. With an A / B stereophony of 20 cm distance between the microphones (Rode NT55) and a distance of one meters to the singer, we had the desired result. The sound was different than usual. The voice sounds more open, bigger and more airy.
Interesting are the slight fluctuations of the voice in the panorama, which are only audible through headphones. 
Although I mixed a digitally generated reverb to the original signal later in the mixdown, the sound was different than in our vocal booth..... You never learn ...

audio/video example (x-mas video the "Händel-Blues")

Stefan Noltemeyer

Donnerstag, 7. November 2019

One Master in two different levels

Now our clients will receive their masters in two different maximized versions.
Version 1 is mastered without a noticeable limiter. This version is used for the online platforms. It has a much higher dynamic than version 2. The second variant is maximized as usual, as it is well-defensible and loud.
to 1: Most online music platforms play their songs at a level of -18 to -14 lufs. Broadcasting establishes a level of -23 lufs (R-128 norm). Loudness Units Full Scale is a frequency weighted unit for loudness. The loudness is determined over the entire title. It makes no sense to create a master that is initially maximized to reduce it by at least 6 dB. So we have decided to create masters which are practically without noticeable use of a limiter. It does not depend one litte dB, because the level is also determined by the arrangement of the song. If a title contains quiet passages, then averaged the loud sections may be a bit louder than if the level were relatively consistent. It is important that this master does not influence the sound by limiting or maximizing. We welcome this development.
to 2: The level on a cd is another topic. Depending on the genre, it still gets really loud here. With a cd master we proceed as usual with a level as loud as possible, without "overdoing it". Of course, it may also make sense to use a limiter as a sound-influencing element to deliberately create a sound of "compression". Stefan Noltemeyer

Mittwoch, 11. September 2019

Coasta Cordalis, restoration of analog master tapes and and single vinyls


An employee of Bellaphon Records has dived into the archive to retrieve old audios of Costa Cordalis and daylight. He came with the 1/4 "masterband of his first album "Folkolore Aus Aller Welt" from 1966 and some vinyl singles.
This material now had to be digitized and, if necessary, worked up.
To my astonishment was the sound that came from the 1/4 "band very well, the sound was stereo and there was almost no noise to hear. The entire album was only produced with two acoustic guitars and a lead vocal. With the pegeltone I fixed the azimuth of the machine heads. Then I set the recording level so that the loudest passages were about -10 to -6dB. In total, there should be 13 tracks on the album. To my surprise, there was a bonus track, which was refereed and additional announcements to the individual titles in the appropriate languages ​​of the songs. These takes were played at the end of the tape. I bounced them together after the digitization to the individual songs. In a particular title, the noise in Intro was a bit strong, so at the request of Bellaphon, I ran away from the section in question. For this job I have several tools of iZotope (RX7) and Steinberg available. 
De-noiseing is generally about reducing noise without affecting the sound of the music. My plug-in is available with the Voxengo Gliss Equalizer in combination with a dynamic equalizer. Both components have intervened very little in the program. Adding both tools quickly stopped the noise without affecting the amount of music.

Much more complex was the challenge of optimizing the sound of old single-vinyl records.
 Here was a whole range of imperfections to work on. The sound was mono, contained plenty of disk crackers and sometimes quite unsightly distortions. First of all, I played the vinyls (dry) as they were. When digitizing it was only about the plates properly in an appropriate level to play in the calculator. Again, I kept the level at -10 dB to -6dB. Also for eliminating the cracker software from Wavelab and IZotope is available. Here the RX 7 worked very well. The cracker I could eliminate it. However, it was necessary to individually identify and mark each cracker and calculate or render with the correct threshhold. If I chose the section too long, because two crackers were relatively close to each other, then the tool "killed" the punch of the snare drum because it considered the transients of the drum to be a crackpot.The RX 7 was also able to eliminate distortion on some passages because it has kept the distortion over transients. This worked very well on some passages, in other places a more specific solution was needed. 

I experimented with a multiband compressor in combination with a de-esser. Distortion in most cases affected the voice in the frequency range between 2000 Hz and 4000 Hz. With the compressor, I reduced the critical frequency range by about 2-3 dB. It turned out, however, that this reduction was not enough on the one hand, but on the other hand, with a greater reduction, the voice became noticeably too quiet. Therefore, I did additionally with the de-esser very selectively reduced the tearing frequency. This narrowband processing allowed me a further reduction of 2 to 3 dB. Once again, I've found that a plug-in for a specific application does almost miracles, but does not provide a usable result for another problems. In any case, it is necessary to deal with the matter exactly. Simple "plug and play" tends to destroy the program rather than achieving a good result. I am amazed at what is now possible with modern software.

Stefan Noltemeyer

Mittwoch, 16. August 2017

10 Tips for using an audio compressor in the mixdown.

1. The best audio compressor is the one whose presence is not heard at all. (Apart from exceptions such as side-chain compression in EDM).

2. A compressor does not increses the signal, but it reduces peak levels. Only the Make Up Gain will raise the level.

3. This does not make the loud sections of the music louder but the quiet.

4. The compressor reduces the original dynamics.

5. Problematic are the transients (fast attacks), for example of an acoustic guitar. The transients compressed at first, because they have a relatively high level. This reduces the level, but also the original sound, because transients are very important for the sound.

6. Voices can be easily reduced because they have very few transients. 12 dB and more (recommended ratio 4: 1) are pssible.

7. But with a strong voice compression, the breather and other background noises are much louder.

8. With the side-chain input, the work of the compressor is not controlled by the input signal, but by the signal that is present at the side-chain input.

9. For example, a synthesizer in a mix can be automatically quieter when the voice is applied to the side-chain input. This creates space for the voice and makes a mix lively.

10. In the parallel compression, the compressed signal and the uncompressed signal are interconnected phase-stiff. This way you can increase the low level signal and keep the original transients. (Dry control)

Stefan Noltemeyer

Freitag, 28. Juli 2017

technical anaysis of a produced song

Before I start mastering a song, I listen and analyze its acoustic-technical problems. I separate the overall sound into its components.

First, I check out the low frequencies. 
Are the basses generated by the bass,the bassdrum or both?
Are there any other instruments which sounds below 100 Hz? -
If the Lo End o.k or should I use a high-pass filter above 30 Hz. (This shows me also my spectrum analyzer)
Are there any problems between the bass and the bassdrum?

Next, I check out the mid frequencies between 200 Hz and 3000 Hz (where the "music plays").
Is there any instrument or a voice that is too loud?
Is there a single frequency that is too loud?
Is there any instrument or a voice that should be louder?
Is the kick of the bassdrum (the attack) and snaredrum powerful enough?

What about the presences (between 3000 Hz and 8000 Hz)
Which instruments are still involved in this frequency range?
Is the voice sound present enough?
Are there problems with "S" in the voice?
Are there instruments and overtones that are to strong here?
Is the hihat too loud (classic error in the mixdown)?
Is Snaredrum still there?

Next check out the high-end above 8000 Hz
Which instruments play "top"?
(Again) Are the hihat, ride and crash o.k.?

What about the total sound
Does anything boom in the bass?
Does the title sound too sharp or too dull?
What happens in the side channel, are there phase cancellations (stereo information)?
What is the total level, has the sum ever been compressed?
Do we need more loudness
if I´m unsure I will listen to the song in comparsion to other titels same genre.

Stefan Noltemeyer