Data Alphabet Soup
Data Alphabet Soup
My last post started with a panicked call about a flash drive. As I worked through it, I realized there was a lot of detail I needed to put on the back burner. One of the significant points I glossed over was the concept of SD card speeds. The difficulty of that discussion lies in the alphabet soup of the transmission and storage of data. In this post, I want to try to answer questions like "How fast is fast enough" or "How big a drive do I need?" The first step of that is to get a clear picture of how we talk about data. In an earlier post on networking, I discussed bits and bytes. Let's begin with a quick review of that.
The most basic math in the digital world is that it takes 8 bits to make a single byte. It is also worth being aware of the concept of a "word," which is simply the fixed size of a data unit for a specific system—such as the 24-bit word of an audio sample or a 64-bit word in modern computing. Regardless of the word size, the 8:1 ratio remains our fundamental translator.
Generally, the industry uses bits to describe data in motion and bytes to describe data at rest. In other words, your network speed is expressed in bits and your hard drive size is expressed in bytes. The shorthand is easy to miss but vital: a lower case b stands for bits, and an upper case B stands for bytes.
Unfortunately, this isn’t a hard and fast rule. You’ll frequently see storage devices or internal transfer speeds measured in MegaBytes per second (MB/s). This is where the alphabet soup gets salty. Because of that 8:1 ratio, a 100 Megabits per second (Mbps) internet connection is actually moving the same amount of data as a 12.5 MegaBytes per second (MB/s) file transfer. If you don't keep track of which unit you're looking at, you might think your gear is underperforming by 800%.
To make matters worse, people—even professionals and manufacturers—are often incredibly careless when writing these abbreviations. You will frequently see a tech spec or a forum post use "Mb" when they mean "MB," or vice versa. This isn't just a typo; in our world, that single character shift changes the value by a factor of eight. When you're trying to determine if a drive can handle a high-track-count session, assuming a Megabit spec is a MegaByte spec can be the difference between a successful recording and a ruined take.
Before we move into the actual math of the media, we should provide a reminder that (once you get to kilo) the metric prefixes (Kilo, Mega, Tera, Peta) each increase by an order of magnitude. We also need to address a specific quirk in how computers use the abbreviations. While "Kilo" means "1,000" (or times 1,000) in the decimal world, the binary world of computers traditionally equates "kilo" to "1,024". Below is a chart with the prefixes and symbols with their corresponding decimal and binary multipliers:
This binary math quirk is why a "2 TB" hard drive appears as roughly "1.8 TB" once you plug it into your computer. The manufacturer is likely using the decimal definition to label the box, while your operating system is likely using the binary definition to report the space. When you are pushing the limits of your storage or bandwidth, that 2.4% difference at the Kilo level compounds into a nearly 10% difference at the Tera level.
There are technical standards (like KiB or MiB) that were created to distinguish between these two math systems, but in the practical world, they are rarely used consistently. You may see a lowercase "k" for decimal and an uppercase "K" for binary, but the inconsistency across manufacturers makes it impractical to rely on capitalization for the prefix. Keep in mind that the same level of apathy regarding the "b" and the "B" can be disastrous. As we established, that distinction is the difference between your session running smoothly or crashing.
Now that we have a foundation for discussing data, we can look a little closer at the amount of data that media actually generates. The resolution (or quality) of most time-based media (as opposed to static images) can be expressed by its bitrate—usually bits per second. The most straightforward example of this is uncompressed Pulse Code Modulation (PCM) audio.
The amount of data that PCM audio generates is a rather simple calculation. Without getting lost in what all the terms mean, the resolution of PCM audio is primarily defined by bit depth and sample frequency. For example, CDs use 16 bits of dynamic range sampled 44,100 times per second. We use the shorthand 16/44.1k to describe this.
The bitrate is simply multiplying the bit depth by the sample frequency:
16 bits × 44,100 samples/sec = 705,600 bits per second (mono)
Since we have two channels for stereo, the calculation becomes:
705,600 bits/sec × 2 channels = 1,411,200 bits per second (stereo)
To calculate the bitrate using the binary metric units we established earlier, we divide that total by 1,024 to get approximately 1,378 Kbps. Dividing by 1,024 again results in approximately 1.35 Megabits per second (Mbps).
For a mono channel of 24-bit/48k audio, the result is 1.10 Mbps. Scaling up to 24-bit/96k stereo, a common resolution for Blu-ray audio, the stream is 4.39 Mbps. For high-end 32-bit/192k audio, the bitrate reaches 5.86 Mbps per channel.
When discussing USB and hard drive transfer speeds, the distinction between moving and static data breaks down. A common configuration for USB 2.0 audio interfaces is 32 channels of 24-bit/48k audio, which generates a total stream of 35.20 Mbps. By dividing by 8 bits per byte, we get 4.4 MegaBytes per second (MB/s). Using MB/s is far more convenient for calculating storage needs. An hour of 32 channels of 24-bit/48k audio is 15.47 Gigabytes (4.4 MB/s × 3,600 seconds ÷ 1,024 MB per GB).
It is also a helpful rule of thumb to remember that for a standard 24-bit/48k stereo pair, you are looking at approximately one gigabyte per hour of storage space.
To put that 4.4 MB/s requirement in context, we can look at the bus that carries it. USB 2.0 has a theoretical maximum speed of 480 Mbps or a theoretical 60 MB/s. However, because of the "alphabet soup" of protocol overhead and the way USB 2.0 manages two-way communication, the real-world sustainable throughput is often closer to 30 or 40 MB/s. Even with that reduced ceiling, our 4.4 MB/s audio stream only occupies about 10% to 15% of the available bandwidth, which is why USB 2.0 remains a reliable standard for multi-channel audio tracking.
While PCM is the standard for professional recording and delivery, we rarely see it in consumer ingestion. In distribution, we almost never utilize the full 1.41 Mbps required for a stereo CD. Instead, lossy compression is employed to discard data that the human ear is less likely to perceive. MP3 was the early standard-bearer for this, but modern codecs like AAC, Ogg Vorbis, and Opus have refined the process, allowing us to reach much lower target bitrates while maintaining high subjective quality. There are even codecs such as FLAC that have “lossless” compression versions relying entirely on data compression rather than psychoacoustics.
A familiar example might be a high-quality 320k MP3. In this context, the compression ratio is roughly 4.5:1 compared to the CD standard. The data is no longer a fixed product of the sample rate and bit depth; it is a variable stream optimized for storage and streaming bandwidth.
If we look at the raw numbers for uncompressed video, we quickly realize why compression is the norm in video. A single frame of 1080p at an 8-bit color depth (1920 x 1080 x 8) contains over 16 million bits of information. If we capture that at 30 frames per second, the uncompressed data rate exceeds 1.4 Gbps.
While we still utilize uncompressed video for baseband transmission over Serial Digital Interface (SDI) in broadcast environments, the bandwidth requirements for recording or over-the-top (OTT) delivery make it untenable elsewhere. Scaling those requirements to 4K (3840 x 2160 pixels) at 60 frames per second pushes the requirement toward 12 Gbps.
Because of these astronomical figures, a video file is rarely a single block of data. Instead, it is a container—like an .MP4 or .MOV—that manages distinct video, audio, and metadata streams, each of which can be compressed differently. The efficiency and final bitrate of that file depend on the codec and settings used. For instance, an H.264 "Long GOP" codec might prioritize Temporal Compression (only capturing changes between frames) for a small file size, while a ProRes codec might use Intra-frame compression (treating every frame as a standalone image) for higher quality and easier editing.
The tool that performs this compression is the codec (short for coder-decoder). Each codec is designed with a standard target compression rate in mind, often expressed in Mbps for streaming or MB/s for high-end recording. For example, a 1080p YouTube upload might target 8 Mbps, a rate that fits comfortably into crowded consumer internet speeds. Conversely, a professional ProRes 422 HQ file at the same resolution targets roughly 220 Mbps (27.5 MB/s), which is better suited for professional network and storage transfer speeds.
These vastly different approaches to compression explain why video bitrates vary so wildly—and how we can almost get to explaining the variety of SD card classifications. To understand how we arrived at these modern flash storage standards, we need to take one more minor historical detour to look at the history of the storage interface versus the drive speed itself.
In the era of mechanical spinning disks, the bottleneck was often the drive's physical rotational speed. A 5400 RPM drive was the baseline for laptops and basic office work. However, 7200 RPM drives became the requirement for servers and high-end media editing because of their faster seek times and higher sustained data rates. Even with these faster platters, the connection to the computer had to evolve to keep up. The older IDE (Parallel ATA) interface eventually gave way to SATA (Serial ATA).
This is where the marketing gets tricky. SATA III is typically expressed as an interface speed of 6 Gbps (Gigabits per second). If we divide those 6,000 Megabits by 8 bits per byte, we get a theoretical 750 MB/s. In reality, due to protocol overhead, the interface maxes out at 600 MB/s. But even that number is just the speed of the interface (the pipe); the physical sustained speed of a 7200 RPM hard drive (the bucket) usually tops out at a sustained write of only 150 MB/s.
When Solid State Drives (SSDs) finally became readily available, we finally started to saturate that 600 MB/s pipe. Now, with NVMe drives using the PCIe bus, we are seeing interface speeds in the thousands of Megabits, with matching drive write speeds in the thousands of MegaBytes.
As we discussed in my flash drives post, the minimum sustained speed matters more to media creators than the marketing metrics of maximum speed. Manufacturers also recognized that we need to clearly differentiate between the interface speed and the media's actual write performance—and to remember that read performance (how fast you can offload your data) is often significantly higher than write performance (how fast you record it). This finally brings us to the evolution of SD card ratings.
First, it is important to understand the physical formats and storage capacities. While standard SD cards and MicroSD cards are physically different, their internal technology is identical. You also see suffixes like SDHC (High Capacity) or SDXC (Extended Capacity), which primarily indicate the maximum storage size and the file system used.
When it comes to speed, the original standard was the Speed Class (indicated by a number inside a circle), ranging from Class 2 up to Class 10. A Class 10 card was the standard for years, guaranteeing a minimum sustained write speed of 10 MB/s. Since our multi-channel audio interface session from earlier only required 4.4 MB/s, a Class 10 card is still more than sufficient for most high-track-count audio recording tasks.
As technology advanced, we had to differentiate between the pipe and the bucket again. Roman numerals like UHS-I or UHS-II indicate the Bus Speed (the interface or pipe). UHS-I is the older, single-row pin interface, while UHS-II adds a second row of pins to significantly widen the data path.
The rating inside the "U" graphic (UHS Speed Class) refers to the memory speed (the bucket) within that interface. A U1 card still guarantees 10 MB/s, while a U3 card guarantees a minimum sustained write speed of 30 MB/s.
Once we reached the high bitrates required for modern 4K and 8K workflows, the U3 rating hit its limit. This introduced the Video Speed Class (indicated by a V followed by a number). In this system, the V number simply indicates the minimum sustained MegaBytes per second. A V30 card guarantees 30 MB/s, V60 guarantees 60 MB/s, and so on.
It is important to remember that while these V ratings are expressed in MB/s (MegaBytes), recording codecs are very often expressed in target Mbps (Megabits). These targets include significant variations in bit depth and compression settings that you have to account for when choosing a card. If you ever forget which symbol is which, a quick search for an SD card speed chart will reveal plenty of visual guides to help you map these buckets to your specific recording needs.
I had a friend once tell me he could tell how big a nerd someone was by the number of letters they strung together in an explanation. He called it the "Nerd Consonant Quotient." In case your eyes have glazed over between all the math in our alphabet soup, here are the high points. If it's a bit confusing, grab some paper and a calculator and go back and try to re-calculate some of the PCM bit rates for yourself.
PCM Calculation: Multiplying bit-depth x sample rate x channel count gives you the uncompressed bitrate. (e.g. 16 bits x 44,100 samples/sec x 2 channels = 1,411,200 bits per second for CD Audio - don’t forget to divide by 1024 for kbits)
Bits (b) are for data in motion (network/streaming); Bytes (B) are for data at rest (storage).
There are 8 bits in 1 Byte. Divide by 8 to get your storage needs from your bitrate.
Computers use 1,024 for metric prefixes (Kilo, Mega, Giga), while manufacturers often use 1,000, leading to that "missing" space on your drive.
Interface speed (the pipe) is the maximum capacity of the connection; Sustained speed (the bucket) is what the physical media can actually handle.
For modern video, ignore the "Max Speed" on the card and look for the Video Speed Class (V30, V60, V90) to ensure your recording won't crash.
Comments
Post a Comment