The resolution of an LPCM recording is determined by the sample rate (how many times per second samples are taken) and the bit depth (how many bits are used
107 KB – 10 Pages
PAGE – 1 ============
Apple Digital Masters: Music as the Artist and Sound Engineer Intended Whether you are a major label, an indie, an aggregator, a recording artist, or an audio engineer, you provide the most important ingredient for Apple Music and the iTunes Store Ñ the music itself. It is our job to faithfully and accurately deliver this music to fans around the world exactly as you intend it to be heard. WeÕve designed work ßows and protocols to facilitate the best possible results, ones that live up to your highest standards for audio quality. To achieve this transparency, you can use tools and technologies from Apple to ensure delivery of the highest-quality master recordings possible into our ecosystem. With more than a billion active iOS devices, Mac computers, Apple TVs and HomePods in the world, thereÕs never been a better time to re Þne, codify, and communicate updated information to you about the best work ßows and processes to produce the billions of Þles streamed and delivered daily to our mutual customers in over 100 countries around the world. Note: Apple Digital Masters replaces Mastered for iTunes as the name of the program to re ßect the fact that these audio advantages are available to our entire music catalog across the Apple ecosystem, whether streamed or downloaded. All Mastered for iTunes releases are now badged as Apple Digital Masters and the Mastered for iTunes software tools are still usable for Apple Digital Masters creation. Innovation and Excellence in Sound Apple celebrates a rich history and tradition of innovation and excellence in sound for computing as well as content creation. The original Mac was engineered fully capable of supporting audio without additional hardware or software, making it one of the Þrst personal computers ever to ship with sound. In fact, Apple received a GRAMMY Award¨ for technical excellence in music, the Þrst and only such award ever given to a personal computing company. When iTunes launched, the decision was made to standardize on AAC instead of the more popular MP3 format simply because AAC clearly provides superior audio quality compared to other codecs at similar bit rates. We worked with Dolby and Fraunhofer and have constantly iterated the Apple AAC encoder to reach the industry-leading quality it achieves today. And today Apple Music delivers lossless versions of these masters to our subscribers using the Apple Lossless Audio CODEC (ALAC.) Following the guidelines and protocols outlined in this document to create and audition AAC encodes, you can achieve simply amazing results: dynamic range thatÕs superior to red book CD audio, and a Þnal product thatÕs virtually indistinguishable from the original studio recording. Technology Brie fApple Digital Masters
PAGE – 2 ============
Mastering for Digital Delivery For decades, the standard for consumer digital audio was the compact disc (CD) and most mastering was done with CDs in mind. But with billions of streams and downloads encoded as AAC to date worldwide, AAC is now the standard for digital music. It only makes sense to create masters speci Þcally for this format. What is AAC? AAC (Advanced Audio Coding) is a format for compressing and encoding digital audio. AAC achieves the portability and convenience of compressed and encoded digital audio while retaining audio quality thatÕs indistinguishable from much larger digital Þles. The iTunes catalog was initially o !ered in 2003 as 128 kbps AAC Þles, many of which were encoded from original CD masters. They sounded greatÑin fact, these downloads led the industry in sound quality. More than 100 million songs were sold in this format in a little over a year, changing the landscape of legal digital music forever. But innovation didnÕt stop there. We have iterated on our encoder for over Þfteen years. Our AAC encoder is now able to transparently encode higher resolution audioÑcreating Þles that retain the small footprint, portability, and ease of use Apple is known for. (And they sound amazing.) High-Resolution Digital Recording Digital audio, such as that on a CD, generally uses Linear Pulse Code Modulation (LPCM Ñoften referred to simply as PCM) to represent audio signals. LPCM works by taking snapshots of the analog audio signal and assigning each a numerical value. The resolution of an LPCM recording is determined by the sample rate (how many times per second samples are taken) and the bit depth (how many bits are used to represent each sample). Higher sample rates can capture higher frequencies, and higher bit depths can accurately represent a greater dynamic range. The standard for CDs is 16-bit 44.1kHz resolution, meaning that the analog signal is sampled 44,100 times per second and each sample is given a value between -32,768 and 32,767. This resolution is often referred to as 44/16. The Nyquist sampling theorem states that to accurately represent a signal, one must use a sampling rate double that of the highest frequency being represented. The highest frequency audible to humans is around 20kHz; therefore a sampling rate of over 40kHz is required to accurately capture the audible range of frequencies. Compact discsÕ 44.1kHz rate is adequate for this need. Even so, many experts feel that using higher sample rates during production provides better-quality audio and a superior listening experience in the end product. For this reason, higher sample rates of 48, 88.2, 96 and even 192 kHz have become standard. In addition, 24-bit recording has become the normÑincreasing workable headroom and total dynamic range. Challenges with Encoding Higher-Resolution Recording An inherent challenge of working with high-resolution audio has been that both the sample rate and the bit depth must be reduced to match the speci Þcations used in mainstream distribution, whether for CD or AAC. This can be done either with software (Òin the boxÓ) or by running the mix through an additional analog stage and resampling. There are pros and cons with each technique, but both can add noise and/or distortion. Downsampling, as the word implies, is the process of using Sample Rate Conversion Technology Brief 2Apple Digital Masters Frequency is the number of vibrations per second and is measured in Hertz (Hz). Human hearing spans a range from roughly 20Hz to 20kHz. Bit rate is di!erent than bit depth. Bit rate indicates how much data is being used per second and is calculated using the sample rate and bit depth. A 256 kbps AAC Þle is encoded with a target bit rate of 256 kilobits/second. It utilizes Variable Bit Rate (VBR) encoding, which uses each bit strategically, dynamically allocating less data for simple sections and more data for complex passages. Dynamic Range , when used as a general audio term, refers to the range of possible volumes Ñthe di!erence between the softest and the loudest parts. In digital audio, aliasing refers to audible artifacts created when higher frequencies are sampled at an insu”cient rate. The result is distortion. A visual metaphor for aliasing can be found in the Òwagon wheel e !ectÓÑa rapidly spinning wagon wheel Þlmed at a low frame rate can appear to be moving backward.
PAGE – 3 ============
(SRC) to lower the sample rate (for example, from 96kHz to 48kHz). This process may create aliasing, an undesired e !ect. Dithering (or adding dither) is a technique used when reducing a ÞleÕs bit depth (for example, from 24-bit to 16-bit). It is an attempt to reduce the distortion inherent in this process. Dither is a tradeo !Ñthe distortion is reduced at the cost of added noise. (Another bit depth reduction option is called Òtruncating,Ó where the additional bits are simply removed. This can cause quantization distortion.) Improved Conversion and Encoding for AAC AppleÕs latest encoding methodology is a two-step process. The Þrst step in the encoding path is to use state-of-the-art mastering-quality Sample Rate Conversion (SRC) to resample the master Þle to a production sample rate. This SRC outputs a 32-bit ßoating-point Þle which can preserve values that might otherwise fall outside of the permitted amplitude range. This Þle is saved in a CAF container. This critical intermediary step prevents any aliasing or clipping that could otherwise occur in SRC. It is this 32-bit ßoating Þle thatÕs used as the input to the encoderÑthis is one key reason we can achieve superior results. Our encoders then use every bit of resolution available, preserving all the dynamic range of the 24-bit source Þle and eliminating the need for adding dither. The advantage of this is twofold. Not only does it obviate the need of adding dither noise, it also lets the encoders work more e “ciently as they donÕt need to waste resources encoding this unwanted and unnecessary noise. By using this highly accurate Þle directly from our SRC and taking advantage of its clean signal, our encoder can deliver the Þnal product exactly as the artist and sound engineers intended it to sound. Lossless Distribution In addition, Apple Music can now deliver lossless encodings to customers using the Apple Lossless Audio CODEC (ALAC.) Best Practices for Apple Digital Masters Our latest high-resolution encoding process ensures that your music is transparently and faithfully distributed in the way you intended it to be heard. However, before you submit songs to Apple for encoding, there are some best practices you can follow to ensure that your audio is optimized for Apple Music or the iTunes store. Use High-Resolution Sources To take best advantage of our latest encoders, use only 24-bit sources and send us the highest-resolution master Þle possible, appropriate to the medium and the project. DonÕt upsample Þles to a higher resolution than their original format. Upsampling wonÕt recover or add information to an audio Þle. Similarly, donÕt Òbit-padÓ or recapture 16-bit Þles in 24-bit. Provide High-Resolution Masters Some mastering engineers prefer to control the SRC process by sending already converted Þles, however we ask that you deliver the highest native sample rate available. As technology advances and bandwidth, storage, battery life, and processor power increase, keeping the highest-resolution masters available in our systems allows for full advantage of future improvements to your or your clientÕs music. Also, though it may not be apparent because there may not always be a physical, tangible master created in LP or CD format, the Apple Music and iTunes catalog forms an important part of the worldÕs historical and cultural record. These masters matter. Technology Brief 3Apple Digital Masters Clipping is a form of audio distortion and can be caused in many ways. In general, it is the result of the amplitude of a signal becoming too great to be accurately represented by a system. In an ampliÞer, this can occur if one attempts too much ampli ÞcationÑthe top of the signal is cut o !, or Òclipped.Ó In digital audio, this can occur when a signal falls outside of an allowed bit- depth range.
PAGE – 4 ============
Master for AppleÕs AAC Encoder When creating a master, mastering engineers take into account the limitations and characteristics of the medium or destination format, as well as the listening environment of their audience. For example, a master created for vinyl is unlikely to be listened to in an airplane or car, and therefore is often mastered for a listening environment where a listener can hear and appreciate a wider dynamic range. Similarly, a master created for a club environment might take into account the noise ßoor of the dance ßoor. Because 256 kbps AAC is a highly portable format, its Þles have the potential to be listened to in a wide range of di !erent settings. So, while one listener may be using AirPods while riding in a loud subway car, another may wind up listening intently to a Bach cantata on a HomePod in their bedroomÑor on a Denon receiver using AirPlay 2 in a home media room. A college student may be deep into Miles Davis’ Sketches of Spain while sporting Beats Studio headphones in the campus library. YouÕre being provided with all the tools youÕll need to encode your masters using the same software that Apple Music and the iTunes Store does so that you can audition exactly what theyÕll sound like to your listeners. Loudness, Dynamic Range, and Clipping Whether youÕre mastering a whisper-quiet zen ßute tone poem or a heavy metal guitar fest, volume and loudness are key issues. The main tools used in masteringÑ equalization, compression, limiting, or combinations of theseÑare all di !erent ways of controlling aspects of volume. Making decisions about gain levels, dynamic range, and frequency response is what mastering is all about. Analog masters traditionally have volume levels set as high as possible, just shy of oversaturation, to improve the signal-to-noise ratio (SNR). With digital masters, the goal is to achieve the highest gain possible without losing information about the original Þle due to clipping. With digital Þles, thereÕs a limit to how loud you can make a track: 0dBFS. Trying to increase a trackÕs overall loudness beyond this point results in distortion caused by clipping and a loss in dynamic range. The quietest parts of a song increase in volume, yet the louder parts donÕt gain loudness and can distort due to the upper limits of the digital format. The advent of digital look-ahead ÒbrickwallÓ limiters have made it possible to increase perceived volume even further. Some artists and producers feel that louder is better. The trend for louder music has resulted in both ardent fans of high volumes and backlash from audiophiles, a controversy known as Òthe loudness wars.Ó While some feel that overly loud mastering ruins music by not giving it room to breathe, others feel that the aesthetic of loudness can be an appropriate artistic choice for particular songs or albums. A less obvious issue when setting gain for digital masters can occur on playback. Whether it’s a compressed Þle like an AAC Þle or an uncompressed Þle such as a CD, digital data goes through several processes to be converted to an analog signal for playback. One common process is called oversampling. This upsamples the digital data at four times the original sample rate to improve the quality of the digital audio signal being converted to analog. If the original digital audio data is at 0dBFS, oversampling can result in undesirable clipping. And if the original was already clipped, oversampling can make it worse. Our recommendation is to leave at least 1 dB of headroom in order to avoid such clipping. In both digital and analog, the highest possible level will vary from track to track, Technology Brief 4Apple Digital Masters Equalization (EQ) controls the volume of speci Þc frequenciesÑsuch as bass and treble. There are many types of EQ, all of which control the volume of speci Þc frequencies. Compression automatically controls volume over time. A compressor decreases dynamic range by decreasing loud signals, increasing lower signals, or by doing both. (Audio compression shouldnÕt be confused with data compression, which is used to reduce digital Þle sizes.) Limiting is a fast-acting form of audio compression with a high ratioÑitÕs often used to attenuate peak levels of a signal that could otherwise result in distortion. By raising the total volume and limiting the peaks, the total dynamic range is compressed, increasing loudness. Multi-band Compression is a combination of EQ and compression where a signal is divided into bands of frequencies which can then be isolated and compressed without a !ecting the other frequencies. For example, one can compress the bass without a!ecting the vocal.
PAGE – 5 ============
depending on the material being mastered. Your decision about the volume and loudness of your tracks is a technical and creative choice. You might decide to take the listener on a dynamic journey through an album as a complete work, raising and lowering the volume level across the sequence of tracks to increase the musicÕs emotional impact. Alternately, you might pursue the loudest possible signal at all times. Whatever you decideÑexquisitely overdriven and loud or exquisitely nuanced and tastefulÑwe will be sure to encode it and reproduce it accurately. We ask that you avoid clipping the signal, and we have provided tools to make it easier than ever to provide clean masters that result in distortion-free streamed and downloaded music. Loudness Normalization Sound Check is a feature available in Apple Music and iTunes that lets listeners hear all their songs at approximately the same loudness. It Þrst determines the loudness of a track and then stores that information in the ÞleÕs metadata. The metadata is then used to raise or lower the volume of each track to prevent jarring volume changes while a device is shu #ing songs. Many radio and streaming services use technology similar to Sound Check to control changes in loudness when playing songs on the air. MP3s have a similar technology to control volume changes, called Replay Gain. The International Telecommunication Union (ITU) is taking a similar approach in its standardization of the characterization of volume in its broadcasting standards (speci Þcally BS. 1770). (Sound Check can also correctly set the volume per album, rather than per song, allowing albums that rely on volume di !erences between tracks, such as The Dark Side of the Moon by Pink Floyd, to maintain their intended volumes.) The e !ect of Sound Check, as well as other volume-controlling technologies, is that songs that have been mastered loud will be played back at a lower volume which can make tracks actually sound weaker. Because many such technologies are available to listeners, you should always mix and master your tracks in a way that captures your intended sound, regardless of playback volume. Remastering and Apple Digital Masters When the CD was Þrst becoming a popular format, many older recordings were rushed to market. To make these recordings available as soon as possible, corners were sometimes cut. In several instances, the wrong master tapes were used or the CD was mastered poorly. Many of these mistakes have since been identi Þed and corrected, but even so, a number of record labels are remastering older material for rerelease in high- resolution formats like DVD-A or SACD. Advances in AAC encoding provide an opportunity for anyone maintaining a catalog of older content to create a digital archive or improve an existing one. There have also been advances in digital mastering tools and techniques. In addition to higher resolutions, other technologies (such as noise reduction and pop and click removal) have undergone signi Þcant improvements. Now that thereÕs a clear distribution path that can take full advantage of high-resolution masters, an older recording may merit being given new life. Delicate and artful remastering can let works of genius be experienced as they were meant to be and haven’t been for years. When remastering for Apple Digital Masters, procedures and best practices similar to mastering for new releases should be followed. You should always work from the best available master, carefully monitor gain to avoid clipping, and audition the encoding process. Technology Brief 5Apple Digital Masters 0dBFS (or ÒZero Decibels relative to Full ScaleÓ) refers to the maximum possible values that can be represented by a digital signal. Values that are beyond the maximum value cause clipping. For example, in a 16-bit Þle the highest value that can be represented is 32,767, and the lowest value that can be represented is -32,768. If you attempt to represent a series of values above 32,767, they will all be clipped at 32,767 and will create artifacts that sound unpleasant and dissonant when played back.
PAGE – 6 ============
Although itÕs possible to remaster from a previously mastered CD source with positive results, in order to qualify as an Apple Digital Master, remastered content must begin with a high-resolution digitization of the original analog source and must sound noticeably superior to the previously released version. Songs and albums submitted to Apple as remastered content will be reviewed to ensure that the sound quality shows discernible improvement. Mastering Tools Mastering is a series of creative choices. As a professional, you can be con Þdent that the work you put in to mixing and mastering your material will be treated just as carefully by Apple. Our role is to faithfully and transparently reproduce the audio you deliverÑ and to make your job easier, you are being provided access to the same software used to encode the entire catalog for the Apple Music and iTunes Library. Whether you opt to use an automated droplet or a command-line utility, the following tools will help you create the intended sound for your masters: ¥Mastering Droplet . The mastering droplet (ÒApple Digital Masters DropletÓ) is a simple, standalone drag-and-drop tool that can be used to quickly and easily encode your masters using the Apple AAC encoder. ¥afconvert . The afconvert command-line utility can be used to encode your masters into AAC using the same commands Apple uses at ingest. ¥afclip. The afclip command-line utility can be used to check any audio Þle for clipping. This is especially important for checking an encoded Þle rather than a 24- bit masterÑthe reason being that levels that donÕt show overs on PCM can still cause clipping when encoded. ¥AURoundTripAAC Audio Unit . The AURoundTripAAC Audio Unit can be used to compare the original source audio Þle to the encoded versions and can check for clipping, all in realtime. ¥Audio to WAVE Droplet . The Audio to WAVE Droplet automates the creation of audio Þles, in WAVE (Waveform Audio File) format, from any audio Þle (such as MPEG or CAF Þles) natively supported on macOS. This droplet is for A/B testing within a DAW. It should never be used to create Þles for delivery . The following sections include instructions for using these tools to convert audio to 256 kbps AAC using the Apple encoder and checking the resultant encodes for clipping. These tools utilize afconvert, which is a part of the Core Audio frameworks in macOS. Changes made to these tools are incorporated into macOS by means of Software Update. Software Update is the best way to keep your systems up-to-date so that the system youÕre using for audio encoding is the same one used by the Apple. (The encoder from Mac OS X 10.6.8 is the minimum required OS for auditioning Apple Digital Masters.) Note that it is possible that the version of the encoder being used by Apple Music and iTunes to encode tracks might not be the current shipping version of macOS. This can result in small digital di !erences in the encodes, but doesnÕt a !ect the Þnal sonic quality or levels that will clip the encoder. Apple Digital Masters Droplet You can use the Apple Digital Masters Droplet to automate the creation of 256 kbps AAC encodes. The Droplet creates an AAC audio Þle from an AIFF or WAVE source Þle by Þrst generating a CAF (Core Audio File) rendered with a sound check pro Þle applied to Technology Brief 6Apple Digital Masters
PAGE – 8 ============
Previewing the Converted Audio File The AAC-encoded M4A Þle (Þnal.m4a) created by afconvert can be previewed in any application that plays M4A Þles (such as the Music App or QuickTime Player). However, if you want to do a more detailed examination (such as loop sections, compare sections with the original, and so forth), you can decode the AAC-encoded data back to uncompressed LPCM data, just as itÕs decoded during playback. You can use afconvert to decode the AAC data to uncompressed LPCM and store it to a WAV Þle. This WAV Þle should only be used for making a more detailed comparison or evaluation beyond what you can do by just previewing the Þle. The deliverable for Apple Digital Masters is the original 24 bit PCM Þle. Use afconvert to decode the AAC-encoded data, as it will correctly decode the AAC Þle, taking care of some details that other applications donÕt do correctly. For example, the resulting decode.wav Þle will have the same number of audio samples as the original source.wav Þle that you started with, and thus it can be exactly aligned with your source Þle for critical listening, comparison, and evaluation. Using afconvert to Decode Your AAC-Encoded M4A File Use the following command to decode your AAC-encoded M4A Þle. In this command, the M4A Þle with AAC-encoded data that was originally generated using afconvert is assumed to be called Þnal.m4a; the WAV Þle generated from decoding the AAC- encoded data is called decode.wav. 1.In Terminal, type on one line: afconvert final.m4a decode.wav -d LEI24 -f WAVE afclip You can use another simple command-line tool, afclip, to check any Þle for clipping. This tool works by examining an audio Þle and identifying areas where clipping has occurred. It accepts audio Þles as input, and outputs a stereo sound Þle containing the left channel of the original Þle and a right channel with graphically represented impulses corresponding to each clipped sample in the original. This sound Þle can then be loaded into a digital audio workstation (DAW), such as Logic, providing a visual map to locate any clipping that may have occurred. The following image is an example of an audio Þle that has been loaded into a DAW to help visually locate clipping.
PAGE – 9 ============
For more information about afclip, type afclip -h on the command line in the Terminal application on Mac OS X. Using afclip to Check for Clipping To check an audio Þle for clipping with afclip: 1.Open a Terminal window. 2.In the Terminal window, type the following on one line, followed by a space: afclip 3.Drag and drop the audio Þle you wish to check onto the Terminal window. 4.Press Return to run afclip. By default, afclip will give a readout of any clipping found. If thereÕs clipping, it will also output a .wav Þle in the active directory. Reading the afclip Readout If any clipping is found, a readout will be printed to the terminal window with details for each instance of clipping. The readout will contain the following information for each instance of clipping: ¥Seconds. The time, in seconds, where the clipping occurs. ¥Sample. The sample number that was clipped. ¥Channel. The channel of the clipped sample. A value of 0 means the clipping occurred on the left channel, while a value of 1 means the clipping occurred on the right channel. ¥Value. The raw value of the clipped sample. Since clipping happens when a value exceeds the range of -1 through 1, these values will be below -1 or above 1. ¥Decibels. The number of decibels by which the sample exceeds the clipping point. The readout will end with a summary of how many total clipped samples the audio Þle contains for the left channel and the right channel. Some readouts will show sample values with decimals, such as 8216364.50 in the table above. This indicates that the waveform will clip between two samples of the original
PAGE – 10 ============
audio data. This is called inter-sample clipping. Most DACs will upsample audio while converting the digital signal to an analog signal, which is known as the reconstruction process. This process can result in clipped values that arenÕt actually clipped in the original material, causing an uncompressed digital signal to cause clipping in playback even if all the values in the waveform donÕt show on-sample clipping. For this reason, itÕs important to check for inter-sample clipping. By default, afclip does inter-sample clip detection and includes the total number of inter-sample clipping at the bottom of the readout, next to the total number of on-sample clipping. AURoundTripAAC Audio Unit The AURoundTripAAC Audio Unit lets you compare audio encoded using 256 kbps AAC against a source audio Þle in realtime. It also includes clip and peak detection, as well as a simple listening test environment. The audio unit can be used in any audio unit host application, such as Logic or AU Lab. Audio to WAVE Droplets The Audio to WAVE Droplet automates the creation of WAVE (Waveform Audio File) Þles from any audio Þle (such as MPEG or CAF Þles) natively supported on macOS. For example, the Droplet can be used to decode AAC Þles to 24-bit WAVE Þles, automating the decoding process used by the command line utility, afconvert, and allowing you to decode multiple Þles at once. To use the Audio to WAVE Droplet, drag and drop source audio Þles, or folders containing those Þles, onto the droplet. The droplet will then convert those Þles to WAVE format Þles. The created WAVE Þles will be titled using the names of their corresponding source Þles, and will be placed in the same folder as the source Þles. NOTE : The Þles created with these droplets should be used for auditioning and comparison purposes only. The deliverable for Apple Digital Masters is the original 24 bit PCM Þle. Addendum: Lossless Audio and Apple Digital Masters Lossless Audio Apple Music now o !ers subscribers lossless streaming and downloads using the Apple Lossless Audio CODEC (ALAC.) There is no need to send two mastersÑthe same Apple Digital Master is used for the ALAC encodes as for the AAC encodes. © 2012-2021 Apple Computer, Inc. All rights reserved. Apple, the Apple logo, Apple Music, iPod, iTunes, Mac, Macintosh, HomePod, AirPods and macOS are trademarks of Apple Computer, Inc., registered in the U.S. and other countries. iTunes Music Store is a service mark of Apple Computer, Inc., registered in the U.S. and other countries. Other product and company names mentioned herein may be trademarks of their respective companies. Product speci Þcations are subject to change without notice. This material is provided for information purposes only; Apple assumes no liability related to its use. April 2021
107 KB – 10 Pages