Before getting started on this task, I spent a little time thinking about issues that could impact the final audio quality and ease of use. These included:
- What format to use to store music files (MP3, WAV, etc)?
- How much disk memory storage would be required for music library?
- What is required to restore the quality of the vinyl record music?
- What are best software tools to capture, process, rip, and burn audio files?
- Best music player software to use?
- Best way to integrate music server with standard audio system?
The following sections provide information that address each of these topics.
Music Storage
If someone is going to invest the time and effort digitizing albums and creating a music server, measures should be taken to insure that the audio quality of each digitized music file is as good as possible. To achieve optimal results with the minimum amount of effort, the goal should be to try get the equivalent to normal CD music quality. And surprisingly, after applying music restoration techniques, the audio quality will be comparable to a music CD in most cases.
CD music is digitized at 16-bits and at a sample rate of 44.1 kHz for each channel. Even though the final bit resolution will be 16 bits, initially digitizing each channel to 24 bits is recommended. This allows all audio restoration data processing to be done at the higher resolution, which will maximize the dynamic range and the final audio quality. Once audio processing is completed, the music file will be down-converted to a 16-bit resolution before it is transferred to the music library.
Given that music is stored using 16 bits at 44.1 kHz, the amount of required storage per minute of raw, recorded music is: (16 bits per music sample) * (44100 samples per second) * (2 channels) / (8 bits/byte) = 176.4 kB/second or approximately 10 M bytes of storage per minute of recorded music. Yes, this is a lot of memory but when you consider that disk memory costs much less than $0.20 per GB, the cost of storing digital music is no longer a cost issue. So the next question is whether music data should utilize data compression or should music data be stored in a raw (e.g. WAV) format to preserve the maximum audio quality? The more popular music file formats, such as MP3 and AAC, reduce music storage requirements about a factor of 10 to 12 at the expense of somewhat degraded music quality. Since a music server application does not have to be concerned with memory costs or data transfer rates, lossy data compression schemes should really be avoided. If it's desirable to have music available in both multiple formats, it's best to choose a lossless format that preserves the audio quality for the primary music archive. Then if necessary, convert files to a more compressed format using conversion tools such as dBpoweramp. When using a lossless compression format, such as WMA lossless, FLAC, or ACC lossless, the approximate reduction in memory storage requirements will be about a factor of two. So using a lossless compression format, a complete album will require about 200 to 250 MB of disk memory storage per album.
Music Recording and Processing
After evaluating numerous software programs to record and process record music, I ended up using Sony Media Sound Forge. If someone is just starting to think about digitizing records or tapes and is looking for a cheap way to get familiar with this process, I would recommend downloading Audacity, which is a free audio editor/recorder. But once anyone gets really serious about digitizing a significantly sized record library and wants to achieve the best audio quality, Sound Forge is definitely the way to go. Although at first the cost of this software may at first seem a bit high (~$250), the quality and capabilities of this software package is clearly worth it. And considering the amount of time that must be invested to record and process each album, the amortized cost of converting each album of a large music library is probably going to be between $1 and $2, again not a high price to pay for the achieving the best audio quality. The Sound Forge package actually includes three critical tools that will be required. First is the recording software which supports the desired 44.1 kHz sampling rate. It also, when used with a high quality sound card, supports digitizing resolutions of 24 bits, whereas other lower cost software programs are limited to 16 bit conversions. Next Sound Forge supports a large set of audio process functions that can be used to optimize audio quality. One of the most important music restoration modules that are included is the Noise Reduction Plug-In Module. When purchased as a stand-alone DirectX module, the Noise Reduction modulesells for over $200, so having it included with Sound Forge is a real bargain. Finally, purchasing Sound Forge includes another application called CD Architect, which can be used to separate the processed music file into individual music tracks and burn audio CD’s. Although the Sound Forge documentation is very good, I would also recommend the Sound Forge Official Guide by Scott Garrigus to gain a better understanding how to use all of the Sound Forge features.
The second key element for music recording is integrating a high performance sound card that provides the hardware to accept an analog audio signal from a stereo amplifier and digitize each stereo channel at 24 bit resolution at a rate of 44.1 kHz. This same sound card will be used in the music server PC to convert digital music data to high quality audio signals that will be played through the audio amplifier. A number of manufactures offer high quality PCI-compatible sound cards. Based upon a review of the M-Audio Revolution 5.1 and a comparison of some other sound cards, I chose M-Audio card and have been very satisfied. This review also recommended a slightly more expensive ($150 vs. $90) Juli sound card from ESI, which has slightly better audio specifications. Although I have no complaints regarding the Revolution card, I would probably choose the Juli card the next time around, not only for the slightly better audio quality but also because this board offers standard RCA audio jacks for all input and output signals instead of mini-audio jacks that are used on the M-Audio and most other sound cards.
Music Players
When I started thinking about digitizing my record collection and creating a digital music library, it wasn’t clear to me what was the best way to play the digital music. Before I decided to configure a music server using a PC, I considered buying a music server from Escient, Olive, and Cambridge Audio. About the same time, I started looking at Windows Media Player 11(WMP 11). From an aesthetics perspective, I think WMP 11 provides a more refined user interface than other commercial music server products that are currently being offered. Like earlier versions of WMP and other popular music players, the music library entries can be grouped by various categories (e.g. artist, album, genre, etc) but unlike other players, WMP 11 displays the original album art with the other music metadata. I personally feel that the integration of album art makes WMP 11 a more visually appealing solution than other options and is also quite helpful when searching for music. So once I decided to use WMP 11, a PC-based music server became the obvious solution.
Music Server Configuration and Integration
If a PC is going to be used as the platform for the music server, what is the minimum configuration to support this function? Since the processing and display demands for a PC dedicated for a music server application are reasonably low, a relatively low cost PC can be used as long as some basic needs are met.
When I started creating my music server, Windows Vista had not been released. And now that it is out, I feel that Windows XP is still the preferred choice since it supports all of the needs of a music server with less memory and processing demands (i.e. cheaper PC). Additionally, when I configured my music server, Vista drivers were not available for any of the suggested sound cards.
In my configuration, a Pentium 4, 2.8 MHz CPU is being used for the music server. Microsoft recommends that the minimum CPU frequency be 1.5 GHz. With a Pentium 4, 2.8 MHz CPU, the CPU load when using WMP 11 to play music is never more than 5%, so CPU demands are relatively low for this application. If you plan to also use the server PC for music digitization and restoration, you want to make sure there is sufficient CPU performance to adequately support the music restoration processing. Using my Pentium 4 configuration, each phase of music processing (several phases are required for each record), required 1 to 3 minutes. So if the CPU is too slow, the processing of audio data may get a bit tedious as you digitize your entire record collection.
Although as little as 128 MB of RAM could be used to support WMP 11, Microsoft recommends that 512 MB of RAM be used. And again, if Sound Forge is going to be installed on the music server PC to process the records, 512 MB is the minimum RAM configuration that should be used.
The amount of required disk storage will depend primarily on the size of the music library. Using lossless compression, a good rule of thumb should assume 4 albums (i.e. ~ 45 minutes of music) per 1 GB of disk memory. So an 80 GB disk should support a music library up to 320 albums. So in most music libraries, a hard disk in the range of 80 GB to 160 GB should be sufficient.
A PCI slot will be required to accommodate the high quality sound card. One additional slot may be useful to support a wireless network adapter if you chose to network the music server with other home PC’s.
The remaining hardware issues to resolve are associated with user interface support. In my configuration, the music server is located adjacent to my flat panel TV. So in order to avoid configuring a separate monitor to view the WMP user interface, I decided to use the TV to support the monitor function (see picture in Introduction). Since most flat panel TV’s support a PC monitor interface this interface is straightforward. In most cases listening to music and watching TV do not occur at the same time so this configuration works quite well. A wireless mouse provides a convenient way to control music selection and playing functions. But be aware that the maximum distance between the wireless mouse and the USB receiver is limited to about 5 feet.