Digital Recorders - Comparison I
Until now, we have supported you in your purchase decisions with our tests for single devices. However, they vaguely hint at the differences in quality of recording devices. With our old tests, a direct comparison could hardly be made, because each device was tested under different circumstances. Therefore, we decided to test all digital recorders at the same time and under similar circumstances, so that you can easily compare them. Additionally, we prepared some sound samples for you. We hope that you enjoy reading our test and listening to the sound samples. Hopefully we can support you in your purchase decision!
We chose five different settings for our recording tests. Those are exemplary for the manifold application areas of digital recorders, and cover a wide range of possible circumstances. The five different settings are an interview with a single person, an interview with a small group, choral singing practice, background noises at a train station in Marburg (Germany) and in a forest. We recorded those five settings with all digital recording devices available, and tried to maintain the same distance and alignment towards the sound source.
Every recording device offers a large set of options, ranging from data compression to the sensitivity of the microphone. In order to compare the recording devices as accurately as possible, we decided to use the same settings for each of them:
- Highest recording quality possible; Ideally uncompressed WAV (24 Bit, 96 KHZ)
- No additional filters or equalizers (noise suppression off, low-cut off, limiter off, etc.)
- The sensitivity of the microphone was adjusted according to the setting (interview with a single person: low sensitivity; interview with a group and choral singing practice: stereo and medial sensitivity; atmosphere: stereo and maximized sensitivity)
- The data was uncompressed and normalized with the software audacity. Then, they were converted into 320 KbPs MP3s (with Lame Version 1.32, Engine 3.97), and can now be found as sound samples on our website. Normalizing sound files is not ideal in terms of sound quality, though, because it can lead to artefacts and noise. However, normalizing the sound files allowed us to compare the sound files with a similar volume, and to examine the characteristics of the individual devices more clearly.
In order to test the voice quality and internal noise of the devices, we had to decide on a device to act as a referee. However, we decided not to use mobile equipment and chose a fixed diaphragm condenser studio microphone instead. We chose a low-noise version worth less than 1,000 Euros: RODE NT2000. We connected it to the pre-amplifier SPL Goldmike 9844 which delivered the 48V phantom powering required for such noble high quality recordings. We expected that the RODE NT2000 set-up would guarantee a very high quality, although we knew that it was impossible to use it for recordings of background noises at the train station. Additionally, we ordered a high quality journalist microphone for the recorders which features a XLR connection: the Senheiser MD 421 II which costs about 400 Euros. This allowed us to test the microphone input of the XLR recorders in ordinary circumstances. Overall, we used equipment amounting to 5,000 Euros for this test.
RODE NT2000 am
Most of these devices are available at our shop, and can be bought at favourable conditions.
In this test, we did not assess the looks, functionalities and handling of the devices, because we dealt with most of these characteristics in our individual tests of the digital recorders (Please note: These tests are currently only available in German). Instead, we focused on the following characteristics: voice, perception of ambience, and noise.
The first part of our test involved the use of studio headphones (AKG K271). We started testing by listening to all recordings blindly, i.e. without knowing which recording was done with which device. (Additionally, the extremely similar conditions of the recordings and settings rendered it impossible to guess which device had been used.) Then, we evaluated the situational recordings with the help of a questionnaire, e.g. "Recording Forest 1: Detail (rippling water) was very precise, considerable noise, very clear, broad stereo recording." While listening to the sound samples, we had to realize that our initial idea of marking the devices was flawed. It was too difficult to fairly mark them. Furthermore, marks would not have been as meaningful as one might expect. The nuances and refinements are too detailed, ambience does not automatically equal ambience, stereo sound can proof to be hindering, and so on. Hence, we decided to visualise the overall picture. Which device left the best impression in which setting, and why do we think so? We filtered the three devices which left us with the best impression for every single one of the five settings.
However, the biggest surprise was that our team came up with almost identical ratings, although we evaluated the devices separately and blindly. There were only marginal differences, but a discussion always set the record straight. Hence, we assume that this is a fine quality characteristic.
If you want to retrace the outcome of our tests, we suggest using high quality headphones for listening to the sound samples below. Notebook speakers and PC speakers are insufficient, because they can neither give an account of details, nor highlight smooth differences existing. Just click on the corresponding "Play"-button of a sound sample, if you want to listen to it. You can also download the sound samples as MP3 files, if you would like to.
Near Marburg, we set up the digital recorders on a big stone table which was in a little forest. We were approximately 4 meters away from a fountain, and 20 meters away from the next street. On the table, all recording devices were lined up in the same direction, lying on a foam mat (except for the model Zoom H2, which was standing on its foot).
This recording starts with a ripple of the fountain, as well as noises coming from the street and the forest itself. There is also a pigeon to be heard. While listening to this recording, mind the different positions of the other birds in the background. The quality of each recording can be determined by listening to the fountain's ripple. Which recording succeeds in precisely displaying details, and which recording is the best?
The fountain's rippling, the high-pitched twitter of the birds, the pigeon, and the perception of ambience influenced our evaluation of the "Forest"-recordings. Determining a winner was surprisingly easy. However, the sheer amount of details we had to consider was astounding.
We clearly favour the following devices:
- Tascam DR-100: sparse internal noise, good perception of ambience, clear details
- Olympus DM-550: dynamic, very clear details, broad stereo recording, high-pitched twitters are well audible
- Olympus LS 10: very clear and natural, grave internal noise (which is, however, endurable), pigeon very clear and not unnaturally displaced, and the fountain is also highly true-to-detail
On the other hand, we clearly disliked the following devices:
- Zoom H2: bass-oriented, muffled, broad but strangely unnatural
- Zoom H4n: narrow, muffled, heavy internal noise, the pigeon was hard to sense
- Edirol: chirping hum, almost mono, muffled
Please note: The recordings of the Zoom H4N and Tascam DR07 are a little quieter than the other recordings. Both of them were positioned on the exterior, and both of the recordings were disturbed by a squall. None of the devices was equipped with a wind deflector.
Here, our team lined up in the middle of the concourse. Again, the recording devices were positioned on a foam mat close to the ground level. A train from Frankfurt arrived at Marburg, shortly before we started recording. Hence, you can hear people who want to leave the train station walking by. You can hear the sound of their steps, and you can hear fragments of different conversations. While listening to the sound samples, try to hear the sound of people deploying the handhold of their trolley bags.
We all agreed that the evaluation of the sound samples gathered here proofed to be the toughest one. We found out that the evaluation of these sound samples depends on the weighting of criteria. Was it about the overall impression, and the neat display of nearby buses and cars? Or was it about tiny details, such as the differentiation of voices and the jingle of money? Hence, we had to deal with those characteristics separately.
When it comes to the overall impression, e.g. including the hum of a car, we decided to favour:
- Olympus LS-11: clear with a good representation of low frequencies, hence a little damped and a clear playback of the background noise
- Tascam DR-100: clear, precise, and natural with a good stereo sound
However, when focusing on details, such as the jingle of money, we decided to favour:
- Olympus DM-550: good representation of detail with a good perception of ambience, well-balanced, barely noticeable noise
- Olympus LS-10: clear, slightly bright with noise, but very true-to-detail
- Tascam DR-100: can be seen as "handyman," it is clear, precise, and natural
Again, we clearly disliked the recordings made with the EdirolR09 ("everything's humming") and the Zoom H2 ("muffled and not precise in the representation of details").
The choir of the Elizabeth-church in Marburg kindly allowed us to listen to one of their practice sessions and to record it. The choir practised with about 60 persons in a room which is about 15*15 meters in size. We positioned ourselves in front of the choir with a distance of about 4 meters. Even though the singing was not ready for its premiere yet, it lead to impressive recordings. (Please note that the sound samples of the LS-10 and LS-11 are a little flawed. You can hear sounds of gripping, because we were not cautious enough, during the recording session.)
While listening to the singing of the choir, we quickly realised differences in quality, and hence we quickly sorted out good and poor recordings. We think that the devices which best suit such a situation are:
- Olympus DM-550: clear voices, good separation of voices, "s"-sounds are perspicuous, nearly without disturbing noises, nice bass without being muffled
- Olympus LS-11: very clear without haze, nice broadness, balanced bass and heights, clear perception of ambience
- Olympus LS-10: many heights (however, it does not sound uncomfortable), overall "light," the best representation of the tenor during the final chord, a little over-driven (that was our mistake, though, and not the device's), good and differentiated (except for the bass)
On the other hand, we disliked the recordings made with the Olympus WS (flat, too high), the EdirolR09 (muffled, too narrow), and the Tascam DR-1 (pulpy, soft).
Here, we are back at our bureau, where most of our early sound sample recordings took place. It is a small room, which is about 12 square meters in size, filled with boxes and tables. We set up the devices on one of those tables. Try to focus on the clarity of the voices, and try to figure out if the background noises disturb the recording process in the second half of the sound samples.
We went to the street with our Senheiser MD 421 II (which costs about 400 Euros), because we wanted to catch some original sounds. We attached the Tascam DR-100 (which costs about 430 Euros) to it. We also used a Olympus DM-550 (which costs about 200 Euros) as a cost-saving counterpart. You can find the (quite interesting) "mini-comparison" here:
|Tascam DR-100+Sehnnheiser MD 421 II||Olympus DM-550|
The test series was impressive. Via headphones, we could relatively quickly spot the differences between the good and the bad recordings. But when it came down to finding the winner of the test, we had to realise that this was not going to be an easy decision. Why? Firstly. there is a great amount of different fields of application which a user could pursue. The same devices react differently in different fields of application. Before making your purchase decision, you should ideally think about the field of application a device is suitable for. Based on this knowledge, you can choose the device you personally favour. (Alternatively, you can also call us and we can counsel you over the phone.) The second reasons is much more pleasant. The devices which performed well are extremely close to each other in terms of quality. This means that there is no winner who left a yawning gap behind him. There is a whole field of good recording devices, and some of them are a little better than others.
Keeping this in mind, we decided on two winners. Both devices are well-rounded products, and the recordings are of high quality. Our winners are the recording devices Olympus LS-11 and Tascam DR-100. Both are of very high quality and suitable for most situations. Deciding on which functions and connection possibilities you want (or have) to use is up to you. Either way, both of them are excellent devices capable of brilliant recordings.
Winner(s) (Overall): Olympus LS-11 & Tascam DR-100
Furthermore, we were surprised by the quality of the recordings made with the model Olympus DM-550. The quality of recorded language is astounding, but the DM-550 is lacking when it comes to the brilliance of music and atmosphere (unlike our two winners). Its price is what makes it special. The Olympus DM-550 costs less than 200 Euros - that is less than half as much money as you would have to pay for the Olympus LS-11 or the Tascam DR-100. Thus, the DM-550 clearly deserves a special price as the "Value for Money"-winner of our test.
Winner ("Value for Money"): Olympus DM-550
Material suitable for broadcasting
Broadcasting - this, surely, is the most challenging setting for recording devices. In search for reliable information, we turned to a public radio broadcasting station. With the information they provided, we determined the general framework. They use a Marantz PMD660 (which should be obsolete by now) with a Senheiser MD 421 II, amongst other things. Furthermore, they use one of the following settings for interviews: mono recording with a microphone (MP3 / 64 kb/s / 48 kHz), stereo recordings (MP3 / 128 kb/s / 48 kHz), stereo recordings with LINE input (MP3 / 128 kb/s / 48 kHz). The standard format for radio broadcasting (digital schedule and archive system) is: MPEG1 Layer2, Stereo, 48.0 kHz, 384 kBit/s. This means that the standard recordings of journalists are recorded with only 64 kb/s (mono) and eventually boosted to 384 kb/s with the studio's system. In this format, contributions are cut and prepared for broadcasting, before they air.
That is good, because every recording device we used is able to produce the formats required. Therefore, there are only few reasons not to use single recording devices for the fields of journalism and radio broadcasting, if the sound quality of the device tested approximates or equals the quality of a recorder with an external Senheiser MD 421 II.
Here, it is important to record voices as clearly as possible. Mono recordings deliver the best results. The recording device should be small and not attract too much attention, so that it can be placed on a table during the recording session. It is turned on at the beginning of an interview, and turned of at the end of it. Comfortable is, of course, a device that produces little noise. However, a little noise does not disturb the transcription progress too much.
Group with four to fifteen people
Recording interviews with a group usually requires stereo resolutions. There is no other way to highlight and differentiate the spatial conditions of the different speakers while listening to the recording. This alleviates the transcription progress. The addition of further microphones (placed in a distance of about 5 meters from each other) can proof to be useful for rather big groups (consisting of 15 or more people), especially if they are sitting far away from each other.
Here, it is important that the recording device is able to rewind frequently, because it is likely that you will want to correct and re-record some passages (edit function and slide switch). Only one speaker who is close to the microphone is to be recorded. The files have to be very small in size, because you might want to send them to a typist via e-mail. Therefore, the sensitivity of the microphone and the quality of the recording have to be adjusted. However, dictation machines are usually not suitable for interview situations. You can find a list of devices which perform well here (Please note: This list is currently only available in German).
Interviews with a telephone
Interviews with a telephone require an adapter (Please note: This text is currently only available in German). Such an adapter is usually added between the telephone and the telephone receiver. You can usually connect every common recording device with an 3.5mm jack plug to an adapter. More expensive (expect a four-digit number price label) are complete telephone set-ups with recording functionalities.
Events and conferences
The recording devices offered here are usually not suited for conferences with 50 or more people. Such settings require technical equipment designed for conferences. However, you can still produce recordings if you want to use them for note-taking later on.
Outlining the own role and position in a transparent manner is important for every scientific investigation setting. This is the only way to enable the reader to evaluate our findings. Dr Thorsten Dresign and Thorsten Pehl are educational scientists and not audio engineers. Our core ability is the recording of dialogue-situations for scientific analysis of the gist of discussion. We have been recording and transcribing conversations for years. Something we regularly do is listening to recordings with headphones. We think that better recordings, i.e. clear and realistic, alleviate the transcription progress, and render it easier to familiarise with a voice. Furthermore, our colleague has been working in the field of journalism (creating coverages of sports, travelling, and nature - things which require clear recordings of nature and music, etc.) for a long time. Hence, he tested the clarity and natural appeal of recordings. We deliberately abstained from analysis supported by frequency curves and noise clearance. Deciding whether a recording is good or not is made with our eyes closed and headphones on.
Last but not least, we earn our daily bread with the sale of digital recorders. Hence, we are not absolutely irrespective. Our goal is, however, to represent our products in a manner as transparent as possible, and to offer well-founded information independent of advertisement and pure technical details. If you want to support us, we would be happy about your purchase of digital recorders in our shop.